here's my prediction of how its gonna play out

AI could end scarcity, end humanity - or boost trend growth by 0.2 percentage points

rewatching 28 days later and impressed by how gentle and humanistic it is

lotta people on here acting like stochastic parrots... responding to new input by regurgitating their training data

Whether you believe #AI is a #bubble or not, what is true is that no bubble in history has ever been discussed so intensely before it crashes. People have usually believed it's up forever. I wonder if that could influence the outcome.

The US government should subsidize Open AI rather than OpenAI

Man I love the Digitakt so much. I aspire to design tools like this… simple features that you can compose to unlock almost infinite creative possibilities…

preview of the near future

I've been noticing people use the word "move" in an interesting way lately e.g. "mental move" or "running a good workshops is one of the most effective moves for making things happen in the world." bengoldhaber.substack.com/p/running-sm...

it really cannot be overstated how good a tacit knowledge database is "is the best practice to do it like this, or like this?" unlocks soooo much in any domain

for me, coming around on LLMs for learning and coding was less "i was wrong, these things are 100% accurate!" and more "wow, I really underestimated how many things are trivially fact-checkable but hard to surface"

One incredible thing about LLMs is that they encode a bunch of best-practice knowledge you would otherwise only pick up through apprenticeship, or by spending months digging through blogs and docs. any% speedruns in new domains are much faster now.

From the standpoint of requisite variety, it is best to let the AI drive vs trying to orchestrate it. Power it up with capabilities and let it sense-plan-act on its own. This is bullish for the foundation model companies...

Solar is fundamentally decentralized. You don't have to wait for state capacity or resolve collective action problems to build it out. And in many places it's not even competing with baseload, it's competing with a dysfunctional centralized grid, or nothing climatedrift.substack.com/p/why-solarp...

subcultures are society buying real options on other ways of being

this web component is now five years old and it has outlived many epochs of JavaScript frameworks incredible journey-ing their “best practices” into even bester practices

Signpost. Been betting for a while that the first truly autonomous agent would be a virus (just like in meatspace).

Wikipedia's "welcome, newcomers" page, April 2001 (just a few months after Wikipedia launched). https://cybercultural.com/p/internet-2001/

Start from: Who am I? What do I know? Who do I know?

i am happy

Batteries are just gonna win. So many avenues for improvement. EV's and appliances will put several TWh of ~free storage on the grid. Density/cost will improve enough for electric trucks, cargo ships, and short-range flying cars.

if you hated the view from nowhere, just wait until you experience the views from somewhere

AI encyclopedias will low-end disrupt Wikipedia for the same reasons that Wikipedia low-end disrupted Brittanica. So the question becomes… how do ground them to prevent them from becoming propaganda outlets (the default outcome)?

for spooky season this year I wrote about Italo Calvino's remarkable 1967 lecture "Cybernetics and Ghosts", wherein he speculates about whether a machine could ever write great literature. joelgustafson.com/posts/2025-1...

I think the cybernetics deserves deep reflection… it offers a way of seeing that resolves many of the dilemmas of dualism, essentialism, etc

mod lists for mod lists 😈

“If you desire to see, learn how to act” - Heinz von Foerster’s Aesthetical Imperative

"That idea is so damn nonsensical that I'm willing to stand on the bridge of a battleship while that nitwit tries to hit it from the air." - Secretary of War, Newton Baker, 1910, on the suggestion that airplanes might be able to sink battleships with bombs

"Risk is when you know the possible outcomes, you just don't know the probability of the outcomes. Uncertainty is when you don't even know the outcomes."

something just hitting me while reading this — openai & anthropic aren’t public companies, but the circular investment patterns mean that they have access to investment from the stock market anyway public without the downsides

Great article. I put the chances of a productive bubble higher simply because I always bet on Perez. This one feels extremely Perezian… bubble in the sense of railroad mania or the telecom bubble. peterwildeford.substack.com/p/ai-is-prob...

effectual mindset vs ineffectual mindset... who would win

it is important to summon the courage to seek out an understanding of what is likely to happen, not just what you hope will happen, or what you think should happen.

Human-AI centaurs is going to be the transition where I stop really understanding folks younger than me I suspect

Been reading Ken Liu’s All That We See or Seem and there’s a lot of interesting near future AI tech. He explicitly connects personal AIs to tools for thought and Vannevar Bush’s Memex, which I think is right.

Happy 34th Birthday HTML tags! On October 29, 1991, the internet pioneer, Tim Berners-Lee, published a document entitled HTML Tags. #WebDesignHistory

Every time I see physical attacks on TEEs, I wish someone would put out a position paper on @hdevalence.bsky.social 's "guy with a glock" model x.com/hdevalence/s...

We take it for granted today, but a single 6W embedded NPU can do more tensor operations per second than a peasant in the 1400s could perform in his whole lifetime

Novelty search!

this follows from a MUCH more general principle

decades of Moore’s Law means there’s plenty of room at the bottom

Once upon a time there was a literary society. In that kingdom, the people loved letters, and they fashioned everything from words. Status, money, knowledge, religion, even the very foundations of the kingdom were made out of words. Then one day a sorcerer came with a magical word generator...

Yep yep yep 💯, i like the active vs passive language. In security circles “trust on first use” is seen as a very bad thing, but strikes me as exactly the right thing here: spend as little of the user’s active trust as possible: by delegating to a secure, passive, zero trust protocol

Truly unfortunate name, but also *technically* accurate. Trust is a scarce resource, limited by Dunbar's number. To increase social trust generally, we must reduce the active trust required to perform day-to-day operations by placing trust in proxies with mechanisms that make defection difficult.

Great quotes here: "No rule of law means high transaction costs." "[When] you cannot rely on the courts to litigate, contracts don't actually matter that much. So therefore you have to do business based on trust. And trust is very expensive."

Note to self: utility classes are an evolutionary dead end. Styles should be component-oriented. Accept duplication and pray for the arrival of mixins.

“Without ‘noise’ there can be no innovation. Total control of the decision-making process at the highest level permits no novelty.” (Gregory Bateson, The Distribution of the Mind)

Bit of a fork from the rest of this thread, but I really like “effectuation” as a way of understanding this approach effectuation.org/the-five-pri...

i don't want them to optimize caves of qud. i want them to go backwards. i want it to take eight minutes to load because it's computing how much a frog hates me

The terminal is the ultimate "Everything" app. It's fun because unlike other apps, customization and user choice is the start. There's no company that owns my PATH, just whatever shit I throw in there and whatever shell I choose sets the tone.