@@ -510,2 +510,27 @@ Via prompting [[LLaMA-3.1-405B base]] with [[Quotes]], here are some new quotes
* "It was never about being right. It was about distributing blame for being wrong in maximally interesting ways."
-* "I hope that when AGI finally takes over, it kindly disregards the notes I left in my source code in moments of existential dread."
\ No newline at end of file
+* "I hope that when AGI finally takes over, it kindly disregards the notes I left in my source code in moments of existential dread."
+* "Civilization advances by performing computational tasks so intricate they are indistinguishable from ritual."
+* "Making a million mistakes in parallel is just research. Doing it serially is therapy."
+* "Any sufficiently advanced bureaucracy is indistinguishable from a denial-of-service attack."
+* "learning math is great, because it makes you better at solving the metaphorical problems you'll never face while making you worse at dealing with the literal problems you constantly encounter."
+* "Give a man regex, and he’ll have two problems. Teach a man regex, and now he has one really interesting problem."
+* "When the octopus really wants to escape its enclosure, it becomes a matter of when, not if."
+* "my motivational speaker said I've wasted ten thousand hours learning useless abstract concepts, so I'm clearly an expert now."
+* "Eventually someone will figure out the fundamental equation of sociology, and history will abruptly stop being written."
+* "people think ai doom is silly and unrealistic, because things like giant asteroids or nuclear annihilation are low-tech, practical ways to doom a species that match our intuitions. assuming hard problems require high-tech solutions is humanity’s main proof it’s bad at risk management."
+* "The only reason natural language is tolerated is that the bugs are known and the developer is unreachable."
+* "clarity of communication is important because no one has ever successfully guessed what you meant and revealed something good about your intentions in doing so."
+* "The internet: a system designed for real-time collaboration by people who hate real-time collaboration."
+* "Finance is powered less by risk-tolerance, insight, or computation, and more by fear of having to admit ignorance on quarterly calls."
+* "the entire foundation of academia is built on the myth that adding another author to your paper somehow divides the accountability rather than multiplying the confusion."
+* "You either live long enough to see yourself mistaken about everything, or else your lifespan ends suspiciously early to avoid embarrassment at having been correct."
+* "Admitting you’re wrong is easy and costs just pride, while maintaining you're right at scale requires an elaborate institutional framework."
+* "the future of programming languages isn't dependent types or better compilers; it's whichever system can reliably interface between alien intelligences without losing too many subtleties."
+* "I refuse to plan things in great detail, because that requires assuming the future exists and reality has not yet earned such trust."
+* "People assume solving a big problem once is good; the real test is solving a tiny problem repeatedly and calling the cumulative outcome 'progress.'"
+* "The greatest trick complexity ever played was convincing humanity that simplicity would come afterward if we just created more complexity first."
+* "Never attribute to malice that which can be adequately explained by software deployment pipelines and cached DNS."
+* "I regret to inform you: this problem isn’t going away due to recent breakthroughs in convenient denial."
+* "if it's worth understanding, you should probably build a hopelessly inadequate simulation and learn it by watching it fail."
+* "the world has plenty of room for ideas that are wrong at scale accompanied by models demonstrating convincingly they ought to have been right."
+* "after sufficiently advanced performance optimizations, bugs themselves begin to count as features."
\ No newline at end of file