catalog
Library
Cards and collections from my Semble library. Pages saved for later, grouped into shelves I return to. 1015 cards, 26 shelfves.
Shelves
26 collections- Collection 9 cards
Modern reading apps
Cool apps to help you read better
- Collection 792 cards
Skyreader Saves
- Collection 2 cards
toread
- Collection 1 card
To process
- Collection 2 cards
Building with agents
- Collection 3 cards
Protocol thinking
- Collection 4 cards
Cybernetics
- Collection 1 card
Cryptocurrency
It's bad
- Collection 2 cards
Thinking about thinking
- Collection 2 cards
ATproto development
Tools and resources for building on atproto
- Collection 20 cards
Cool Atmosphere apps
- Collection 9 cards
Internet sensemaking
- Collection 5 cards
The structure of social media
- Collection 9 cards
Books I've been reading
- Collection 3 cards
Tools for thought
- Collection 1 card
Cool tools
- Collection 2 cards
Awesome terminal
- Collection 8 cards
Local first
- Collection 5 cards
Tech right analysis
- Collection 9 cards
Security?
- Collection 1 card
Tech and Law
- Collection 118 cards
the AI of it all
- Collection 5 cards
understanding events
- Collection 1 card
vc stuff
- Collection 12 cards
development
- Collection 19 cards
atproto stuff
Recently filed
page 6 / 43-
Google Reader was building the wrong future
The app that taught us to directly follow our favorite creators.
-
Permissioned Data Diary 7: Off the Record
In which we put records in a repo, sign them, and sync them (but not quite the way you think).
-
Arvind Narayanan (@aisnakeoil)
Companies check their own work through various internal but independent functional units: QA, security red teams, model risk management in banks. I think it’s time for AI evaluation to become one such unit. Orgs deploying AI should stand up cross-functional eval teams with their own reporting line. Many reasons: 1) Evals as IP / moat. It’s now widely recognized that evals are the new IP. So it makes sense to have teams whose primary focus is on creating and widening this moat. 2) Evals are harder than you think. This is less well recognized but as someone whose research centers on AI evals this has been my consistent experience. It can't be an afterthought and must be a center of excellence. 3) Evals are inherently cross-functional and require a distinct set of skills. They are judgment heavy, require both AI expertise and deep domain expertise, as well as customer understanding and sophisticated thinking about risk. To do them well, you need competence in data science & stats, business operations, product/customer experience, IT, risk management, and even compliance (depending on the sector). 4) In-house but independent eval teams keep companies honest. A climate where teams are getting top-down mandates to hit deployment targets and show results has resulted in a culture of companies fooling themselves. It is extremely easy to knowingly or unknowingly to do evals poorly, making your AI deployment look much more successful than it is. Eval teams who don’t share the deploying teams’ KPIs are the best defense against this.
-
The Resonant Computing Manifesto
Technology should bring out the best in humanity, not the worst—a manifesto for resonant computing built on five principles that reject hyper-scale extraction for human flourishing.
-
AI #177 Part 1: Tip of the Iceberg
This week saw the releases of, among other things:
-
Tinker
Tinker is a training API for researchers and developers.
-
“Useful” is not sufficient
So Linus Torvalds, head of the Linux kernel development, put his foot down on the Linux Kernel development mailing list when someone was bringing up criticism of LLMs: “Linux is not one of those anti-AI projects, and if somebody has issueswith that, they can do the open-source thing and fork it. Or just walk away. […]
-
Monthly Roundup #44: July 2026
It’s a quiet week so let’s do the monthly right on schedule.
-
The Future Worth Building Is Human
AI built for autonomy crowds people out, making us passive observers of what’s coming. We’re building toward a different future.
-
Inkling: Our Open-Weights Model
Our first open-weights model: multimodal, Mixture-of-Experts, with controllable reasoning effort. Available to fine-tune on Tinker.
-
Holdfast 0: Situating
This post introduces the written version of some optimistic and constructive thinking on knowledge in networks that I’ve been wrestling with since the end of last year.
-
In The Imagined Classroom: Hogwarts
Teaching Fictions #2
-
What is “loop engineering?”
There’s talk about loop engineering, but what is it exactly? I looked into it, and found triggers, cron jobs, AI slop & more. Is it a “here today, gone tomorrow” trend?
-
Twitter Thoughts For You
I previously have written back in March 2022 about how I use Twitter, and back in April 2023 about Twitter and its then-new algorithms, which have changed again.
-
The political economy of billionaire derangement
The passions are devouring the interests
-
'love island' and the mirror of red pill ideology in modern dating
red pill masculinity, the objectification of women through entertainment ecosystems—and how it helps script modern heterosexual dating culture
-
Harari vs. Henrich
What science actually says about human evolution
-
‘David Bowie was a crazy workaholic’: Labyrinth at 40 – an oral history
Today, Jim Henson’s dark fairytale is seen as a classic of 80s high camp. But on release, it bombed. Here, members of the cast and crew remember laughter, tricky puppets and Henson’s ‘joyful magic’
-
Chili Peppers of the World: Cultivars, Species, and Heat
An illustrated guide to chili peppers of the world, organized by Capsicum species and cultivar, with notes on origin, heat, form, domestication, and global food history.
-
I wanted to be Anthony Bourdain—until I met him.
CW: suicide, suicidal ideation, addiction
-
Nenex: A Neural Personal Wiki Idea
Proposal for a personal wiki built on neural nets: all edits are logged & used to finetune a NN assistant in realtime.