Issue 34 · October 24, 2026

Nothing Matters More Than Civilization

Written without AI. A stream-of-thought first pass; only obvious typos fixed.

Nothing matters more than civilization. It's not pretty or perfect, but homeless, destitute, addicted, and mentally ill people can survive today in and around major cities like LA. How long did any of us survive before modern civilization? What could our scientists, thinkers, builders, and doers actually accomplish pre-civilization? Elon Musk born in prehistoric times builds what exactly?

So while it's easy to get caught up in our individual lives, careers, politics, and causes, none of this is really anywhere near as important as the flourishing of modern civilization. Without civilization, life is brutal, hellish, and short. Without civilization, human ingenuity makes small gains and spreads slowly.

So maybe it's worth asking what civilization-ending risks, if any, does AI pose? What core infrastructure (internet, communication, investing/banking/finance/financial records, utility, military, government) could AI credibly threaten? My guess is a rogue human/AI shutting down some core infrastructure - like eliminating banking/investment records, shutting down most or all of the internet or power grid - and what happens if you can't access your bank account or those records are completely wiped out? How do hospitals function? Grocery stores? How does the supply chain operate? AI could also accelerate nuclear and biological threats, but pure digital infrastructure destruction seems a more plausible immediate threat.

This raises the question of what is the best AI strategy? Full speed ahead? Try to stay ahead of foreign and open weight models and secure critical infrastructure - in essence, try to always stay one step ahead in perpetuity? Or until some plateau in model capability is reached? Or pause and work on alignment which may or may not be fully realized and may risk being surpassed or at least threatened by foreign/open weight models that aren't necessarily working as hard on alignment or rigorous guardrails?

Without knowing answers and solutions to these questions, it seems like we should be asking "what are our backup plans?". If any critical digital infrastructure gets wiped out/shut down, what are our options? I don't know the answers. Is anyone in a position to take action/develop backup plans even asking these questions? I guess that is my main concern here. Perseverating on pausing vs accelerating AI development misses the more important question of what is our plan if/when something critical breaks? We are focusing on how fast we should go instead of (the more important issue of) how to handle a crash.

This is the author's stream-of-thought first pass, written without AI. Only obvious typos were fixed, and the title was added from its first sentence. It was the raw draft the author gave to the human/AI editorial runs (Muse Wonder, Claude Opus 5.5, Gemini 4 Argon, Kimi K3.1 Pro, and GPT-6 Astra). It was never shown to the model that wrote the machine essay, or to the Stage C critics.