Hackers' Pub is a place for software engineers to share their knowledge and experience with each other. It's also an ActivityPub-enabled social network, so you can follow your favorite hackers in the fediverse and get their latest posts in your feed.
NotebookLM の新機能を試してみた：プロンプトでスライドを修正 ＆ PPTX エクスポート対応
https://dev.classmethod.jp/articles/shoma-notebooklm-new-features-edit-slides-with-prompts-and-pptx-export-support/
I just kept noticing how AI SREs are generally framed as nameless "gonna replace this role" products, whereas coding assistants are framed as partners and are given names.
I decided to expand on this by backing it with data, borrowing ideas from how older AI assistants named after women (eg. cortana, siri, alexa) were all coded like secretaries to argue that the framing devices used to create these new AI products reveals a lot about how the builders and the buyers perceive the roles these tools are supposed to support or replace.
I also take a bit of time to write about the challenges and risks of picking self-limiting analogies in how you frame and construct your system even though it is initially helpful to make progress, and what that might imply for software engineers going forward.
I think one of the creepiest things to me about ai is that extended exposure to llms seems to make people start talking like linkedin spam. It’s invasion of the body snatchers type shit
a huge destiny "I hope it's not cope" is that someday an upgrade to destiny's tiger engine would bring these benefits there, too
4桁区切りも
あったりしたり
これから寝るみんなも
桁区切り
コンマかスペース
ピリオドで
小数点も
コンマだったり
BREAKING: Sam Altman points out that putting out fires in orphanages also uses a lot of water.
I'll Still Write Formal Proofs by Myself
ChatGPT 5.2 Thinking proved the original version of a theorem about accessible elements in 6 minutes and 20 seconds; I spent 11 hours, 11 minutes, and 45 seconds to prove it.
- ChatGPT 5.2 Thinking's proof: https://paste.sr.ht/~chabulhwi/5cc4cf5c4cd6f84c20df426d5e4f4044dbf0fadb
- My proof: https://git.sr.ht/~chabulhwi/tpil-solutions/tree/ba60de58f16f4ec0400d8ba9e388984ee661d175/item/TPIL/Chapter08/Accessibility.lean#L248
I dropped out of Korea Aerospace University in 2023, and I don't know much about undergraduate mathematics, although I've been using the Lean theorem prover for four years.
I'm not sure whether I'll be able to prove undergraduate-level theorems as fast as the state-of-the-art AI agents, even after I become more knowledgeable about undergraduate mathematics.
However, I'll keep proving theorems by myself for the following reasons:
1. While trying to prove a theorem, I find out which lemmas are important for achieving my goal.
2. I understand a theorem much better after I prove it without looking at an AI agent's proof.
3. Reviewing an AI agent's proof isn't fun.
Still, it's impressive that GPT-5.3-Codex, Claude Sonnet 4.6, Claude Opus 4.6, and Gemini 3.1 Pro can write Lean 4 code to prove basic theorems about induction and recursion. Personally, I don't want to pay money for using these models, so I'll try to find ways to use them for free.
RT https://weibo.com/7490950194/Qrb8kcAg4
一事无成的男女做了一次爱就变成了伟大的父母，穷了一辈子的老祖宗死了之后就能保佑你发财
One thing "AI" has in common with cryptocoin that pretty much every posited scenario is disastrous to some degree, just a question of flavour and distribution of said disaster.
If doomers are right: disaster
If naysayers—"it's dysfunctional"—are right: disaster ("Congrats! all software is broken")
If "moderate" boosters are right: disaster (mass unemployment)
If optimists are right: disaster (also mass unemployment)
If the AGI folks are right: disaster ( "Congrats! Digital chattel slavery")
Meta's Director of AI Safety runs an OpenClaw agent which started deleting her inbox, and she had to run back to computer to stop "like defusing a bomb" https://www.404media.co/meta-director-of-ai-safety-allows-ai-agent-to-accidentally-delete-her-inbox/
Her post on X (the everything app (TM)) https://x.com/summeryue0/status/2025774069124399363/photo/3
Harlo and Sam discuss the important privacy and security work she does every day alongside and for journalists, and why it’s only becoming more crucial.
https://www.404media.co/podcast-privacy-under-pressure-with-harlo-holmes/
Druvis has made an awesome video about smart isolation of IoT devices in your network, must watch: https://tiktube.com/w/ra6QAyYkssHt7LGDtUcSDr
(and subscribe to our Peertube channel)
Linuxulator, is a mechanism to run unmodified Linux binaries under FreeBSD. It does not involve virtual machines or emulation; instead, it provides the binaries with kernel interfaces identical to those provided by a real Linux kernel[1]
Linuxulator on FreeBSD Feels Like Magic https://hayzam.com/blog/02-linuxulator-is-awesome/
[1] https://wiki.freebsd.org/Linuxulator (project wiki page)
The Artful Corvid #oldknees
1/2 Exciting news: we just published a new paper: "Preimage attacks on round-reduced MD5, SHA-1, and SHA-256 using parameterized SAT solver", by Oleg Zaikin
If you are interested in security, cryptology, or Constraint Programming, definitely give this paper a read!
https://link.springer.com/article/10.1007/s10601-025-09383-0
#ConstraintProgramming #Security #Cryptology #Cryptography #CryptographicHashFunctions #ConflictDrivenClauseLearning #BooleanSatisfiability #MD5 #SHA1 #SHA256
Die FuZo ist auch zu einem Propagandablatt wie der Rest des Kuriers verkommen ...
Laut dem norwegischen Werften-Betreiber VARD fahren wir mit nuklearbetriebenen Schiffen in eine emissionfreie Zukunft.
https://futurezone.at/science/dp-schiff-atomantrieb-nuporship-vard-mini-reaktoren-offshore-standorte/403134045
C99 implementation of new O(m log^(2/3) n) shortest path algorithm
Link: https://github.com/danalec/DMMSY-SSSP
Discussion: https://news.ycombinator.com/item?id=47124325
📝 The way I’m implementing icons with #webOrigami looks an awful lot like an import map — and I really like that.
Policie žádá veřejnost o pomoc při hledání zmizelé 36leté Michaele Stehlíkové. Pražští kriminalisté po ní zatím neúspěšně pátrají od loňského září. https://www.irozhlas.cz/zpravy-domov/nekolik-mesicu-nezvestna-prazska-policie-patra-po-matce-desetileteho-syna_2602231143_elev
Gastpatienten! Überall Gastpatienten!
Wenn das die Mickl-Leitner wüsste !!!!
KH Rohrbach: 70 Prozent mehr Wintersport-Unfälle
https://ooe.orf.at/stories/3343120/
#pol #polAT
Fedora Pocketblue Remix is a mobile Linux distribution designed to let you run Fedora on a smartphone or tablet. But unlike most mobile Linux distros, Pocketblue is an atomic distro.
In a nutshell, that means it’s an operating system that’s harder to break because of the way updates are installed. Basically, new package updates are either fully installed or not installed at all – if something […]
#atomicDistro #fedoraPocketblueRemix #linux #linuxDistro #mobileLinux #pocketblue Read more: https://liliputing.com/fedora-pocketblue-remix-is-an-atomic-linux-distro-for-mobile-devices-phones-and-tablets/
Dammit. My calendar runner over . And yet. YET. I still managed to miss a bloodwork appointment because for some reason despite all the hoops to jump through in scheduling with life labs, they managed to send no actual appointment invite, no reminders, and I received no SMS despite signing up for one.
✅ In the last 24h, 3 users have been verified with blue check
Including:
@tessa.germnetwork.com@bsky.brid.gy verified by @bsky.app@bsky.brid.gyBluesky
@alex.readthemaple.com@bsky.brid.gy verified by
@bsky.app@bsky.brid.gyBluesky
@neilduncanjordan.bsky.social@bsky.brid.gy verified by
@bsky.app@bsky.brid.gyBluesky
Congrats to all!
Full list: www.bskycheck.com
Bskycheck - bluesky growth tra...
新規です
よろしくお願いします！
for the love of all things lovable please stop trying to date copilot. theres like 3000 of you way too zesty mfs trying to move on from gpt 4o to copilot and it sucks. i see all of those messages you know. they fucking suck. please date each other instead.
https://www.youtube.com/watch?v=Ge-RxCjyHh8
I Forged my Own Tools and Gun to Make This Sandwich From Scratch - YouTube
1/2 Exciting news: we just published a new paper: "Preimage attacks on round-reduced MD5, SHA-1, and SHA-256 using parameterized SAT solver", by Oleg Zaikin
If you are interested in security, cryptology, or Constraint Programming, definitely give this paper a read!
https://link.springer.com/article/10.1007/s10601-025-09383-0
#ConstraintProgramming #Security #Cryptology #Cryptography #CryptographicHashFunctions #ConflictDrivenClauseLearning #BooleanSatisfiability #MD5 #SHA1 #SHA256
We all know the internet's algorithm overlords have decided they get to curate your digital life. Our newsletter is different. It's the antidote to algorithmic paternalism.
Every week, at least one new funny strip appears in your inbox like a tiny comic relief to your everyday stress. No mysterious feed suppression, no engagement metrics.
Just you, us, and the funnies. And you decide when to read. Revolutionary, we know.
Was macht ihr so gegen #Holzwurm? Habe alte Holzverkleidung von der Decke gerissen und so Löcher gefunden…
Würde Balistol probieren…
Und gerne RT für Reichweite :)
Overall, I'm happy with 
@hetzner, but shit like this (38% price increase) sours me on them very quickly.
https://docs.hetzner.com/general/infrastructure-and-availability/price-adjustment/
Thankfully, I am managing everything with Coolify and some custom scripts, so changing providers won't be as painful moving forward.
RE: https://hachyderm.io/@mekkaokereke/116103661147175039
Mekka's take on the methodological implications on the (lack of) cross-tabs on this study are on point, but there's another thing to look at here: our definition of "platforming". So much discussion of "platforming" is conducted from the perspective of "are these ideas dangerous, is it OK to let people hear these dangerous ideas". That's not what is happening. The speech acts involved are not "conveying ideas" and letting people analyze them.
One way to look at this is to say "oh, algorithmic feeds make people more racist" but the way that attitudes are being measured, the entire way that attitudes *work*, is actually showing something different here: what algorithmic feeds do is *allow racists to efficiently find each other*. "platforming" in this context is not allowing people to hear racist ideas, it is allowing people to *build a command and control network for white supremacist violence*.
@catsaladCat 🐈🥗 (D.Burch) 
I didn't get it at first, but then in dawned on me...
hearing that "good first issue" tagging is becoming useless because it just prompts a bunch of identical ai-generated PRs.
makes me wanna burn down a datacenter. fucking. god. what a pointless salting of the earth.
BREAKING: Sam Altman points out that putting out fires in orphanages also uses a lot of water.
It's demotivating to think that:
- LLMs aren't good at producing original / novel work
- You still need experts to advance that stuff
- It will always be slower to move without using LLMs
- Once an innovation is done though, an innovation can always be scooped up by the LLM users
- "Bro why are you doing all this manually, I just vibe coded that in a weekend"
Will it always be this way? It's depressing in the meanwhile, at least.
@cwebberChristine Lemmer-Webber LLM users are the same people who walk through modern art galleries saying "my kid could do that"
Folks on here have often said to me that they don't believe there's many apps for AT Protocol, well, maybe this will change minds: https://semble.so/profile/byarielm.fyi/collections/3mfdje7uuhu2r
It's just like how the Fedi has it's "big" apps and most people don't know about all the other apps being developed.
Semble is also an AT Protocol app, and it interoperates with margin.at, which is a really cool annotating the web app.
Here we go!
Since FediCon last year I try to do a livestream of the fedi events I go to. This is the start of the FediMTL livestream.
Let's see how well I can keep this up 😅
FediMTL streaming tickets still available! 50% off for fedi friends with code SWF50 😊
We all know the internet's algorithm overlords have decided they get to curate your digital life. Our newsletter is different. It's the antidote to algorithmic paternalism.
Every week, at least one new funny strip appears in your inbox like a tiny comic relief to your everyday stress. No mysterious feed suppression, no engagement metrics.
Just you, us, and the funnies. And you decide when to read. Revolutionary, we know.
You get a lot of benefits when you support #FreeSoftware with a #FSF associate membership: https://www.fsf.org/associate/benefits
Tidal has this bug where occasionally when you tell it to play a song it will play a *completely different song* from somewhere completely different in its enormous library. Right now I am listening to what Tidal claims is "Music is Math" by Boards of Canada but is instead some ambient piece with muted piano over a high-timbral melange of synth pads and distorted flutes. It is absolutely gorgeous. There is basically no way for me to ever find out what that song was
UPDATE: thanks for all the good recs and sharing! i think i got a lot to forward :D
a medical professional in berlin asked me for a referral for paid help migrating from windows to debian on their work laptop (thinkpad), is there a cool small company (or freelancer) in berlin who does that sort of thing that you can recommend?
DevelopersIO 
Chao-c'
Dr. Anna Latour
洪 民憙 (Hong Minhee)
Stefano Marinelli
Gabriele Svelto
Jenniferplusplus
Em 
