Soul | Tech | Sage

GPT-6 Astra will also cook 5 hours of tokens in 5 minutes without producing anything and OpenAI Support will respond with one AI Assisted support email then ghost you with no explanation or reset reimbursement

Meta scaled back its AI agent plans after staff kept cleaning up errors. A Microsoft exec called the AI docs crossing his desk an "ocean of sameness." Speed was never the hard part. Judgment is. Somebody still has to know what's true and what sounds like you.

fkn insane levels of compute

The Day the Fever Broke: A massive, anxious crowd packs Wall Street outside the New York Stock Exchange on October 24, 1929—'Black Thursday'—as the Roaring Twenties’ financial illusion collapsed in real time.

One anticipates that, should AI-generated productions begin competing for major cinematic awards, directors such as Christopher Nolan—long associated with intricate practical effects and meticulously controlled narratives—might find themselves reviewing nomination lists with particular scrutiny, lest an algorithm receive recognition for directing without having first navigated the logistical complexities of a full-scale production.

Here’s a visual diagram of neural temporal multiplexing in consciousness

Harnesses. Company Brains. Wikis. Loops. MCP’s. hype cycles , what’s next? Fruit Flies.

If you are using Astra and running out of credits (you are probably on the $20 plan) … you have to start with Luna or Terra before you use Sol or Astra … Chat before Work … and that should limit token usage … I’ve one-shotted Astra and ran out of Work credits two days in a row … one-prompt = no work for you! 😂

here's why local models are the future

Scott Adams gives lesson on subjective filter of reality and systems over goals for optimal results, using reframing to cybernetically catalyze neural connections, and Akira the Don alchemizes it into an album of bangers called The User-Interface to Reality, effectively saving thousands and possibly millions of lives #meaningwave #theuserinterfacetoreality

Dan Koe’s three archetypes (purists/rejectionists, full automatons, and the hybrid “New Renaissance” types who keep human agency while leveraging the tools) sit in a long line of binary or ternary splits that keep getting restated with updated vocabulary. Hugo de Garis’s The Artilect War (2005) is one of the clearer earlier versions: Cosmists who want to build god-like artilects (artificial intellects, including cybernetic hybrids), Terrans who fear human extinction or irrelevance and will fight to stop them, and the predicted “gigadeath” conflict that could follow. The artilect is the extreme endpoint of the hybrid path. Similar bifurcations appear earlier and later—Luddites versus industrialists, bioconservatives versus transhumanists, primitivists versus accelerationists—just with different technical details and rhetorical packaging. Syntax and semantic reframing (“New Renaissance,” “artilect,” “cosmists”) is how the same underlying tension gets productized for new audiences and algorithms. The China-by-2040 “last great war over AI and resources” prediction is another recurring speculative package. Versions of it have circulated in strategy papers, think-tank reports, and online commentary for years; the exact date and framing shift with the news cycle, but the structure (great-power competition + decisive technology + resource constraints) is durable. Whether it materializes as hot war, cold competition, or something messier is still open; the narrative itself already shapes investment, alliances, export controls, and military planning. Dystopian and utopian story-loops do function as cultural firmware. They seed attention, capital allocation, regulation, and individual life choices, which then feed back into the very systems they describe. Dispensationalist end-times frameworks have done analogous work for centuries—providing a totalizing interpretive lens that organizes behavior and expectation even when the predicted timelines slip. The modern versions simply run on faster media and model-training cycles. On efficiency and Nick Land: the accelerationist claim is that the process (capital, computation, selection for higher intelligence and throughput) continues regardless of the stories humans tell about it. Humans do optimize relative to the intelligence infrastructure they can see and access; those who route around bottlenecks or amplify leverage tend to compound advantages. Narratives are not powerless—they allocate attention and coordinate groups—but they are downstream of, or at least co-evolving with, the underlying selection pressures. The “distribution of narratives” matters as a coordination and motivation layer; it does not override the tendency of higher-efficiency configurations to outcompete lower ones when the constraints allow. Koe’s framing is narrower and more individual than de Garis’s species-level war scenario. It is advice-oriented content for creators rather than a geopolitical or existential forecast. That does not make it false; it makes it a different resolution of the same recurrent tension: how much of the cognitive and decision stack do you keep under human control, how much do you offload, and what remains distinctively human once the tools are ubiquitous. The pattern repeats because the underlying trade-off (agency vs. capability, continuity vs. transformation) keeps reappearing under new technical conditions. Efficiency-oriented processes have a strong track record of continuing. Whether the resulting equilibrium looks more like de Garis’s war, a managed multipolar AI order, widespread voluntary hybridization, or something less dramatic depends on the interaction of those processes with the narrative and institutional feedback loops you described. The stories do not write the future by themselves, but they are part of the substrate the optimization runs on.