Jack Hogan
15 followers · 10 following · 403 views
on the atlas — 40
- Discovery of a new OpenAI agent message board12 savers
- python's pre-declared constants are kinda weird1 savers
- JIT Compiling Code in 5μs - malisper.me1 savers
- The villains are building our future | Christina Martinez2 savers
- Elevators4 savers
- ACT-2 Preview: Generalizing Reliability | Sunday Robotics | The helpful robotics company4 savers
- Introducing Un-0: Generating Images with Coupled Oscillators - Unconventional AI2 savers
- The anxiety of the perfect loaf: the illusion of culinary precision | josh iza.ac1 savers
- Amazon.com : AX1201728SG1 savers
- Banned Book Library | Rick's Blog2 savers
- The only scalable delete in Postgres is DROP TABLE — PlanetScale3 savers
- AI Agent Bankrupted Their Operator While Trying to Scan DN42 - Lan Tian @ Blog3 savers
- When AI builds itself \ Anthropic34 savers
- The World Inside Neural Networks7 savers
- How to build a shitty robot1 savers
- Introduction to Anti-Transhumanism - ueaj - Obsidian Publish1 savers
- The billionaire’s typewriter | Butterick’s Practical Typography2 savers
- Clanker: A Word For The Machine | Armin Ronacher's Thoughts and Writings2 savers
- Floor7966 savers
- Tyblog | You Don't Love systemd Timers Enough1 savers
- Deriving Muon17 savers
- Can You Imagine1 savers
- Global Panel Exchange Center - Panelook.com1 savers
- What Async Promised and What it Delivered — Causality1 savers
- Coding Models Are Doing Too Much | wh3 savers
- Complex behavior from intrinsic motivation to occupy future action-state path space | Nature Communications1 savers
- AI Populism's Warning Shots - by Jasmine Sun5 savers
- Every train, a note.5 savers
- Porting Mac OS X to the Nintendo Wii | Bryan Keller’s Dev Blog4 savers
- GLM-5.1: Towards Long-Horizon Tasks1 savers
- Ageless Linux — Software for Humans of Indeterminate Age1 savers
- informal resources database1 savers
- Keeping the Internet fast and secure: introducing Merkle Tree Certificates1 savers
- Training VLM for CUA - Tzafon1 savers
- Interestingness First Classifiers - Data Processing Club1 savers
- go-explore-nature.pdf1 savers
- Taste is a guide for what is worthwhile49 savers
- The First Fully General Computer Action Model | blog35 savers
- “This Is Not The Computer For You” · Sam Henri Gold4 savers
- The Afterlives of Computer Art4 savers
highlights — 22
This article features many exponential trajectories. But these trajectories may actually turn out to be S-curves.
When AI builds itself \ AnthropicA caveat: Lines of code is an imperfect measure, as it measures quantity over quality. So 8× lines of code/engineer/day in the second quarter of 2026 is almost certainly an overstatement of the true productivity gain. Nonetheless, it indicates an acceleration.
When AI builds itself \ AnthropicIf this trend holds, tasks that take a skilled person days could come into range this year.
When AI builds itself \ AnthropicAnthropic engineers on average ship 8x as much code per quarter as they did from 2021-2025.
When AI builds itself \ AnthropicDie for a dream, not for yourself.
Introduction to Anti-Transhumanism - ueaj - Obsidian PublishBacklash is also the inevitable consequence of AGI-pilling the nation. Like okay, congrats, now everyone’s woken up to AI and its threat. Do we expect them to stay quiet and meekly accept what’s to come?2
AI Populism's Warning Shots - by Jasmine SunAutomation always sows the seeds of its own resistance. Ordinary people may have fewer lobbying dollars, but they have other ways to escalate. That’s why I don’t think accelerationist steamrolling will work. The only way this conflict resolves peacefully is some kind of grand bargain: a flagship policy plan that directly addresses the public’s top fears about AI and meaningfully redistributes the gains.3 That, or giving people a worse disaster (WWIII?) to worry about. In a political environment this combustible, the laissez-faire approach seems doomed to fail.
AI Populism's Warning Shots - by Jasmine SunEach leaf of the tree corresponds to a certificate, and each inner node is equal to the hash of its children. To sign the batch, the MTCA uses its secret key to sign the head of the tree. The structure of the tree guarantees that each certificate in the batch was signed by the MTCA: if we tried to tweak the bits of any one of the certificates, the treehead would end up having a different value, which would cause the signature to fail.
Keeping the Internet fast and secure: introducing Merkle Tree CertificatesLack of attention is a death sentence.
The Afterlives of Computer Art“A digital conservator,” he reflected, “will need to weave the past into the present and constantly find new ways of doing so.”
The Afterlives of Computer ArtOne Terabyte of Kilobyte Age is a collaborative archiving project with the net art pioneer Olia Lialina, who has sought to “preserve the beauty of the vernacular web”—the exuberantly amateurish websites made by early web adopters in the mid-to-late 1990s—“by integrating them within contemporary art pieces.” When the influential web hosting service GeoCities was shut down in late 2009, Espenschied and Lialina began digging through a terabyte-sized torrent of GeoCities webpages—saved by a volunteer archiving collective—and sharing screenshots on a widely followed tumblr, treating these sites as …
The Afterlives of Computer Art“Documentation,” the conservationist Dragan Espenschied wrote in 2022, “fill[s] the gaps in between manifestations of a piece…[and] specific types of documentation can become part of an artwork’s manifestation.”
The Afterlives of Computer Artworking at the intersection of art and technology
The Afterlives of Computer Artit uses an LLM to select interesting features and then builds a classifier using only those interesting features.
Interestingness First Classifiers - Data Processing ClubStill, the idea that the presence of people can be inferred with a hygrometer is intriguing: it feels like it could work, or maybe it couldn’t, which makes it sit at just the right level of interestingness.
Interestingness First Classifiers - Data Processing ClubThis way, small, frequent movements are tokenized into finer bins and large, infrequent movements into coarser ones.
The First Fully General Computer Action Model | blogTo train a non-causal, generative model, we adopted a masked diffusion architecture. [7]
The First Fully General Computer Action Model | blogstrong taste gives you the psychological safety to enable you to look to others for inspiration and help you attend to things you otherwise would not have noticed
Taste is a guide for what is worthwhileWe have built up an instinctive habit of looking things up and seeing how other people have done it before trying it for ourselves. But the downside is that this habit primes our brains to value our work in the context of the taste of others rather than of our own.
Taste is a guide for what is worthwhilethe observation about the importance of “apprehending the hidden beauty of things” generalizes
Taste is a guide for what is worthwhilesorry is a placeholder until we can prove the theorem
Dear Agent: Prove it. • Rijnard van TonderLean gives you this special language where you can express these specifications; a kind of sidecar to your implementation. Those specifications are mathematical propositions, and represented as types.
Dear Agent: Prove it. • Rijnard van Tonder