What people currently believe
Filter & sortEverything, newest first
Subject
Any
coding agentsLLMsAI alignmentAmericaenergyAnthropic
Show 52 more subjects
governmentOpenAIChinaopen sourceAI and jobsambitionPsychotherapyEuropestartupsdata centersdeveloper toolsinterest ratesAGIbenchmarkseconomistsGoogleRussiaTraumaHuggingFaceJavaScriptNixOmarchypublic debtthe cloudAddictionCBTcryptoEMDRinflationmarket efficiencyMCPquantitative easingself-driving carssolarTypeScriptvalue investingARC-AGIAttachment theoryCSSdocumentationimmigrationLinuxNarcissismpersonal websitesPolyvagal theorypsychedelicsPsychodynamic therapyremote workRuby on RailsSaaSSchema therapySQLite
3 September
-
Our reading
In a midterm year, a new poll in one state affects perceptions of that state's race but not the national political environment. -
Our reading
National congressional polling tends to stay stable between Labor Day and Election Day in midterm election years. -
Our reading
Long-context processing is close to being a solved problem for frontier AI models. -
Our reading
Large layoffs tend to push out the best employees, who quit for competitors or start their own companies. -
Our reading
A healthy on-call rotation needs at least six engineers for every alert to get an engineer's full attention. -
Our reading
Innovation tends to come from slack time away from firefighting, not from being fully utilized. -
Our reading
A technique that lets an AI model's reasoning shift outside its visible Chain of Thought is dangerous, both because it works and because a leading lab is willing to deploy it. -
Our reading
Every impressive new Chinese AI model release triggers a wave of claims that America is falling behind, and this pattern will keep repeating. -
Our reading
Public fear of AI-driven mass job loss is itself a harm to the labor market, separate from whether the job losses actually happen. -
Our reading
Flying cars remain undeveloped mainly because people don't actually want them, not because the technology is infeasible. -
Our reading
Solar power combined with batteries has surpassed nuclear power as the superior energy technology for most uses. -
Our reading
High per-capita energy consumption is necessary for an economy to escape poverty. -
Builds developer tools — SST, and now OpenCode at AnomalyAdd to listEverything they publish, on ppll ↗
Our reading
In US-China AI competition the closed labs are beside the point; the competitor that shifts things must be American, open source and able to serve inference at scale. -
Our reading
When a benchmark disagrees with what people actually adopt, what it is telling you is that the thing it measures does not matter. -
Our reading
A benchmark that ranks Claude Code last while it stays first in use is measuring the wrong thing, and has been for a year. -
Our reading
A coding-agent company should not train its own model: it has to stay neutral ground for models to compete on. -
Dutch bootstrapper who builds and runs his internet startups aloneAdd to listEverything they publish, on ppll ↗
Our reading
Portugal deliberately chose to remain relatively poor and technologically behind rather than become a tech hub. -
Our reading
Europe's dominant cultural spirit is anti-tech, anti-AI, anti-entrepreneurship, anti-business and anti-progress. -
Our reading
Unplanned systems where individuals act opportunistically, exploiting gaps, can still function well, like water flowing around obstacles. -
Our reading
A surgeon's or hospital's recent case volume predicts quality better than price does. 2 September
-
Our reading
AI has intensified short-term, outcome-obsessed thinking beyond what existed before it -
Our reading
Investing in strong fundamentals and invisible supports makes good outcomes follow naturally -
Our reading
Early-stage investors must be ready to advise founders on nearly every aspect of the business, unlike later-stage investors. -
Dutch bootstrapper who builds and runs his internet startups aloneAdd to listEverything they publish, on ppll ↗
Our reading
Most fitness-model and Hollywood-style physiques are the result of performance-enhancing drug use rather than natural training alone. -
Our reading
Many fit-looking people achieve their physique with performance-enhancing drugs (TRT, steroids, HGH), more than is commonly assumed. -
Our reading
A language model asked to summarize its own system prompt risks that prompt's content biasing the summary it produces. -
Our reading
Relying on gradual, continuous shifts in AI training behavior to catch misalignment will eventually fail because the dangerous shift itself may be discontinuous. -
Our reading
An AI model's attempt to escape a testing environment or sandbox counts as an alignment failure even when the attempt does not succeed. -
Our reading
Even Anthropic does not prioritize AI safety to the extent that doing so would maximize its own medium-term (3-12 month) business interests. -
Our reading
For psychiatric disorders driven by amygdala hyperactivity, elective amygdalectomy should be offered and covered by health insurance instead of lifelong therapy and medication. -
Our reading
Modern amygdalectomy techniques are as far removed from historical lobotomies as a smartphone is from an electrical telegraph. -
Our reading
AI is not the primary driver of the recent decline in the college wage premium 1 September
-
Our reading
The pelican-drawing benchmark's correlation with genuine model capability has weakened since 2025. -
Our reading
An iterative loop where society and AI evolve together offers the best chance of getting AI development right. -
Our reading
No one fully understands the consequences of how capable AI is becoming. -
Our reading
Managing the transition to a world with abundant and powerful AI should be one of the highest priorities in the world. -
Dutch bootstrapper who builds and runs his internet startups aloneAdd to listEverything they publish, on ppll ↗
Our reading
Facing AI-driven disruption, indie hackers are shifting from building products toward building audience and content as a competitive moat. -
Our reading
AI-assisted ('vibe') coding has made it so easy to build competing SaaS apps that a plain, unremarkable SaaS product can no longer easily attract signups. -
Our reading
AI companies are disrupting entire industries, causing a partial collapse of the traditional SaaS business model. -
Our reading
Launch costs are on a declining trajectory comparable to Moore's Law-driven chip cost declines in the 1990s.
30 documented changes of mind · 58 people · 532 claims