Just posted a new piece on medium: . It comes complete with a handful of interactive agent-based models created with a purpose-built framework. I may have gone a bit overboard... 😬
Alex Komoroske (
@komorama
) is one of the most original, articulate, and insightful thinkers I've ever come across.
He spent 13 years at
@Google
, where he led Chrome’s Open Web Platform, spearheaded augmented reality inside Google Maps, and developed toolkits to align
I wrote a quick strategy parable about the tension between getting things done now (builder) vs tending patiently to future possibility (gardener) and the power of discovering the acorns amongst the pebbles:
🚨New
@InfiniteL88ps
is up🚨
Fantastic chat with
@komorama
about complex, adaptive systems, information flow, context switching and luck surface-areas, Saruman vs. Radagast magic, Schelling points and how to escape busyness + much more!
🔥🔥🔥
Are we too fixated on boldness? Join me in reflecting on the importance of recognizing our limitations, the resilience found in fluidity, and wisdom of the tortoise over the hare:
The podcast episode of Clearer Thinking I recorded with
@SpencrGreenberg
is out! We talked about slime molds, serendipity, doorbells in the jungle... you know, the usual.
Delighted to welcome
@DanielaAmodei
(back) to
@Stripe
for AI Day — this evening we'll be chatting about Claude 2 and her vision for Anthropic. Here's a primer: .
I just published this week's Bits and Bobs:
Nanotech cages for circus bears. Frankenstein ecosystems. Aggregator deserts. Knowhow cauldrons. Self-ratcheting quality. Being boxed in by the Himalayas.
Sat down with
@AnthropicAI
cofounder
@DanielaAmodei
at
@Stripe
AI Day: .
We discuss:
-Claude 2 launch
-Lessons from working at Stripe
-Would you rather have Isaac Newton or LLMs?
I just published my weekly reflections:
Enchanted artifacts. Dandelion fields. Gilded turds. Sublime jumbles. Ecosystem surfing. Galaxy-scale intuition.
Just published this week's bits and bobs (my personal reflections):
A few of the topics this week: Names as boxes. LLM-fu. Whale falls. Engagement and co-ownership.
I just published by weekly reflections:
Slop. Continuously viable. Contextual integrity. Leaving a dot unconnected. PhD thesis rabbit holes. Miyazaki-style forest sprites. Jurassic park harmonica.
I just published my weekly reflections:
Accelerated copy paste. Fracking vs precipitating apps. Private vs Confidential Compute. Macro-Maxwell's Demon. Flood-fills of possibility. Malleable vs remixable. Swiss cheese defense. Embassies vs couch surfing.
Posted a quick piece this weekend about how EVERYTHING is a trade-off, and once you acknowledge that it opens up powerful new approaches to surf through all of the complexity.
I just published my weekly reflections:
Ephemeral micro-apps. Jagged intelligence. Dancing with the LLM. Tech islands. Fractal branching. Viral needle-in-the-haystack discoveries.
I just published my weekly reflections:
LLMs as compilers. Context gardening. Glop. Squish-to-fit systems. Ofuda enchantments. Hour-glass architecture. The personalization paradox. Driving a big rig like a sports car. Anxiety transmutation.
I just published my weekly reflections:
Data gravity. Adopting an elephant. Stigmergy swarms. JIT schemas. The game designer mindset applied to LLM software. A whole universe of previously non-viable situated software.
I just published my weekly reflections:
Ofuda. Disposable software. Contextual flow control. Supercritical coffee shops. Primitive archeology. OODA leash.
I just published my weekly reflections:
LLMs as preternaturally patient people. Radagast inside, Saruman outside. Denaturing radioactive data. Systems on the edge of criticality. Parables and emojis in arguments.
I just published this week's reflections:
The same-origin paradigm. Confidential computing. Stars aligning. Lilliputian rope. Hyper-dimensional folds. Idea cauldrons. The heat death of a paradigm.
I'm trying something new: developing nuggets of insight in a collection of public, inter-linked cards in a custom CMS at . This companion twitter account will tweet out a few cards a day, and is the best way to follow along!
Hello there! This is the companion account for , which tweets out a few cards a day. Following this account is the easiest way to dip your toes into the content without getting overwhelmed. Learn more at
I just published my weekly reflections:
Intelligent personal data. App husbandry. LLM literacy. The Walmart of Finance. Commoditized understanding. Copernican trauma. General intelligence.
I just published my weekly reflections:
The ouija board effect. Idea labyrinths disguised as idea mazes. The coasian boundary of software. Software with a market of a single user. Wannabe aggregators. Prosuming.
I just published my weekly reflections: .
Adjacency maps. Meta-ecosystems. The quality that cannot be named. Illegible gauntlets. Safe swarm intelligence.
I just published my weekly reflections:
Lilliputian webs. Silver swarms. Hyperstition. Gauntlet communities. Anti-consensus swarms. The magic of preparation.
@deepfates
@vgr
Other decks I have done (which sadly I no longer have access to) are more explicitly slime-moldy. I even built a little tool to help me build the diagrams for them!
I can't get enough of this absolutely sublime deck
@dglazkov
made to succinctly and intuitively explain the cynefin model of predictable and unpredictable problem spaces.
I just published my weekly reflections:
Required tar pits. Shower-time creativity. Herding butterflies with a screwdriver. Gilded turds. Collection as creation. Cookie monster providers. Joyful slogs.
I just published my weekly reflections:
LLMs as compilers from English to code. o1 as System 2. Enchanted data. The Coasian theory of the app. Spontaneous software. Blackboard systems. Consensus as entropy. Terraforming nerd clubs. Learning swarms.
I just published my week reflections:
Imitation intelligence. Last principles thinking. Hallucinated mini-apps. Psychological safety paradox. The "look what you made me do" dynamic. A goldrush red ocean. Faux typos for "authenticity".
@KaseyKlimes
@callumflack
@CardsCompendium
Yeah agreed I’ve felt that way a number of times. The card format is nice for simplicity (limited length) but other times is too constraining.
I just published by weekly reflections:
Epicycles and curlicues. Invisible mech suits. The Minecraft punching-a-tree to redstone-computer gradient. Niches as protective caves. The peek and poach pattern. LLMs' secret lair problem.
This paper shows that learning about how your firm works is equivalent to gaining Lovecraftian Secret Knowledge. Promising managers asked to do process redesign at their firm learn how messed up it all is, so they give up rising careers & become hermits at the edges of the firm
I just published my weekly reflections:
Crystallized mirrors. The magic of discretionary effort. A butler's loyalty. Boring mistakes. Mimetically-fit stories. The illegibility reaper. LLM's subterranean success. LLM-induced hysteresis.
I just published my weekly reflections: .
Intelligent lorem ipsum. Surfing hidden waves. Performative cathedral playbook. Artfully half-assed. Tip of the sphere. Tangled web of tautologies. From huh? to 'duh! Horses for the mind. Chainmail feedback loops.
@ben_mathes
Ah yes, I think the enlightened and defeatist stance can look superficially similar and people might fall into the latter accidentally. I also think people resist the enlightened form because they erroneously pattern match it to the latter. I added a bit to the post on this.