Publishers killed 12ft Ladder months after Elon made paywall bypassing famous.
In August 2023, Elon Musk told his followers they could read the New York Times for free using a paywall bypass site. That single post pulled in 37 million views.
Back then, the bypass everyone actually built their workflow around was called 12ft. io, nicknamed 12ft Ladder. Paste a locked article's link in, get the readable page out.
It ran on one domain, owned by one company. That made it an easy legal target. In July 2025, the News Media Alliance got it shut down for good.
So a developer rebuilt the same idea with one difference. This time there is no domain to kill.
It's called Ladder, and it lives on GitHub.
You run it yourself. One Docker command spins it up on your own server, under your own name, with your own rules for which sites it touches.
Nobody can pressure a registrar that was never theirs to pressure. There is no single company holding the project hostage, because every person who runs it becomes their own company.
8.5K stars. GPL-3.0 license. 100% open source.
https://t.co/5N0gwHcw9y
I genuinely don't understand why everyone isn't using this yet
Andrej Karpathy, a co-founder of OpenAI, posted a simple idea that hit 16 million views: stop using AI to write code, use it to build a second brain.
You point Claude Code at a folder, drop in any source, an article, a transcript, a PDF, and Claude reads it, links it, and files it into a living wiki of everything you know. It compounds like interest, the more you feed it, the smarter it gets.
Here's the whole thing:
> Install Obsidian, create a vault, open it in Claude Code
> Paste Karpathy's wiki idea file and tell Claude to build it
> Claude makes three folders: raw for sources, wiki for its pages, a CLAUDE.md that runs it
> Drop any source into raw and say "ingest this"
> Ask questions across everything, forever
Five minutes to set up, and you never start from a blank chat again.
Full step-by-step guide with Claude and Obsidian, link below.
Bookmark this
How PatchGuard detects hypervisor-based hooks. Written by two Riot Games Anti-Cheat Engineers. Some of the most underrated researchers in the Windows internals and kernel/hypervisor security space.
Microsoft embedded multiple detection techniques inside PatchGuard to catch hypervisors hiding the real LSTAR MSR value. Trap flags, MOV SS tricks, ICEBP edge cases, LSTAR shadowing, and deliberately triggered page faults that expose the real syscall handler address.
Part 2 includes a custom detection technique with a full working PoC.
Part 1: https://t.co/pJNWLBDuQW
Part 2: https://t.co/nheDEfeDD1
PoC: https://t.co/e5PDrvaOj3
Authors: @nickeverdox@aidankhoury
#WindowsInternals #ReverseEngineering #AntiCheat
Write your Own Virtual Machine
teaches you to build a VM that runs actual assembly programs, like a working 2048 or roguelike.
the final code is ~250 lines of C and runs on the LC-3 architecture
Researcher found a design flaw in EasyAntiCheat's driver that allows injecting and executing unsigned code inside any EAC-protected game process. Even with HVCI and Secure Boot enabled.
Covers how EAC bridges between the kernel driver and the game's modules and how a flaw in that bridge gives unrestricted code execution.
Post is from 2021 but the design patterns in anti-cheat service initialization remain relevant.
Blog: https://t.co/IQpjy8OX4z
Video: https://t.co/1lLzzCx7z2
Authors: @sinclairq8@_xeroxz
#AntiCheat #ReverseEngineering #InfoSec
Un desarrollador chino llamado tw93 se hartó de que sus aplicaciones de escritorio le devoraran la RAM y el disco.
Abría Slack y desaparecían cientos de megabytes. Abría Discord, Notion o cualquier otra app y pasaba lo mismo. ¿La razón? Casi todas son lo mismo por dentro: un sitio web empaquetado con una copia completa del motor de Chrome (Electron).
Decidió que tenía que haber una forma mejor.
En 2022 empezó a construir Pake. Usó Rust + Tauri, que en vez de incluir un navegador completo, aprovecha el WebView nativo del sistema operativo.
El resultado fue brutal:
- Slack con Pake → 8 MB (en vez de 524 MB)
- Discord con Pake → 9 MB (en vez de 265 MB)
- ChatGPT con Pake → 9 MB (en vez de 260 MB)
Cuatro años después, su repositorio tiene más de 51.000 estrellas en GitHub. Tiene builds listos para Grok, ChatGPT, Gemini, Discord, YouTube, Twitter y muchos más. Todo bajo los 10 MB, ligero, rápido y gratis.
Y lo mejor: con un solo comando puedes convertir cualquier página web en una aplicación de escritorio nativa.
No fundó una startup. No levantó inversión. Solo resolvió un problema que molestaba a millones de personas.
A veces el cambio real lo hace una sola persona que se cansa de las cosas como están.
Esta brutal, repo en los comentarios 👇
Incredible how Z. ai literally has their RL infrastructure open source.
The entire OPD post-training of GLM-5.2 took on this slime platform took ~2 days.
https://t.co/XVjW6rGcbg
This 2 hour Stanford lecture shows exactly how Stanford trains it's engineers to build AI systems. It's more practical than every Claude tutorial & prompting threads you've seen.
Bookmark & give it 2 hours, no matter what. It'll be the most productive thing you do this weekend.
@russell_m@MarinaMedvin@Starlink Not if you want to live. There's a reason this uses short bursts over a wide spectrum to evade triangulation. ELINT is a deep subject.
@bear_jackthe@MarinaMedvin The exact schematics can be made public and it won't make it less secure it's assumed that these will captured and reverse engineered so we build things with that in mind. Much like open source code for security applications seeing it doesn't make it any more vulnerable.
@GlowinOwen1982@kimmy22dawn@MarinaMedvin Well take that up with the people who published it and took it off classified information not someone who can use google. They can get access to spy satellites you think they can't use google? Security through obscurity isn't used here, it's why it's not classified.
@kimmy22dawn@MarinaMedvin It's public info on the internet. An army with an intellgence division / someone with a laptop can google everything she said. In fact you can get manuals for it. But knowing how it works won't help it's assumed some people will be caught with these. Doesn't change their efficacy
JUST IN: The last time the CIA ran a deception operation to extract Americans from Iran, it invented a fake Hollywood production company called Studio Six Productions, took out ads in Variety, printed business cards, and walked six diplomats through Mehrabad Airport disguised as a Canadian film crew scouting locations for a science fiction movie. That was January 28, 1980. The operation was called Argo. It took months to build the legend. It worked because the cover story was elaborate enough to survive scrutiny at an airport checkpoint during a revolution.
Forty-six years later, the CIA built the legend in hours. According to Fox News citing a senior Trump administration official, the Agency launched a deception campaign inside Iran on April 3, leaking false information that the downed F-15E weapons systems officer had already been located and was being moved on the ground for exfiltration out of the country. Israel Hayom, citing a senior CIA official, reported that operatives used “fabricated transmissions” to convince Iranian forces that the pilot had already been extracted and moved to various locations. Some reports describe a maritime variant: a ghost ship, a fake naval exfiltration that drew Iranian coastal forces to a vessel that did not carry the man they were looking for.
Iran chased the ghosts. While IRGC ground forces, Basij militia, and Bakhtiari tribesmen pursuing a $60,000 bounty hunted the false leads, the WSO sat in a rock crevice on a 7,000-foot ridgeline in the Zagros Mountains with a pistol, an encrypted beacon, and SERE training, waiting for the people who actually knew where he was. The CIA’s “unique, exquisite capabilities,” the phrase used by the senior official and left deliberately undefined, confirmed his identity, verified the beacon was not a trap, and passed the coordinates to the White House. A senior Trump administration official told Fox: “Within 8 hours we had planes in motion. Within 12 hours we were on the ground in Iran.”
The deception did not achieve total diversion. IRGC forces still converged. A firefight erupted. Two transports were destroyed on the ground by the Americans who flew them there. But the fabricated transmissions bought the hours that mattered: the hours between the WSO being a hunted man alone in a crevice and being a located man with hundreds of operators converging on his position. The fake narratives did not need to fool every Iranian commander. They needed to slow enough of them long enough for the real operation to reach the ridgeline.
Argo took months and extracted six people from an airport. This operation took hours and extracted one man from a mountain 200 miles inside hostile territory during an active war. The tradecraft evolved from elaborate legend-building to rapid tactical disinformation injected into adversary networks in real time. The principle is identical: create a reality the enemy believes long enough for the people you are saving to reach safety.
The man the Iranians were chasing was never where they were looking. The ship they were watching was never carrying him. The convoy they were pursuing did not exist. And by the time they understood what the CIA had done, SEAL Team 6 had already loaded the colonel onto an aircraft and flown him to Kuwait, where the President said he “will be just fine.” The Agency that once used a fake movie to rescue Americans from Iran just used fake radio transmissions to do it again.
https://t.co/dAOBBMsgDS
A Triple Spring-Mass Pendulum Just for The Fun of Physics
We model the mass positions as
r₁ = (x₁, y₁), r₂ = (x₂, y₂), r₃ = (x₃, y₃)
with y positive downward in the simulation. The three spring segments are
r₁ = r₁ − (0,0)
r₁₂ = r₂ − r₁
r₂₃ = r₃ − r₂
Each spring applies a Hooke force along its own direction,
F(r; k, L) = −k (|r| − L) (r / |r|)
and each mass also feels gravity,
Fgᵢ = (0, mᵢ g)
So the equations of motion are
m₁ r̈₁ = Fg₁ + F(r₁; k₁, L₁) − F(r₁₂; k₂, L₂)
m₂ r̈₂ = Fg₂ + F(r₁₂; k₂, L₂) − F(r₂₃; k₃, L₃)
m₃ r̈₃ = Fg₃ + F(r₂₃; k₃, L₃)
We integrate these ODEs forward with RK4 using a small dt. Different scenes come from changing the mass ratios and from giving the third mass a small initial kick.
The yellow trail is the path of the third mass.
The inset phase portraits show the same motion in state space:
(x₃, vₓ₃) and (−y₃, vᵧ₃)
When those curves stay tight and loop cleanly, the motion is more regular. When they spread, fold, and smear, energy is being redistributed across the whole chain between gravity, spring stretch, and kinetic motion.
#NonlinearDynamics #CoupledOscillators #SpringPendulum #ChaosInMotion #ComputationalPhysics #MathematicalVisualization