The jobs threatened most by AI are jobs that rely in information arbitrage. “Consulting” firms will get less business when people can just consult AI. Odds are high that the consultant would just be using AI on the backend anyway. Just skip the middleman. What survives and thrives is physical arbitrage. Access capital, resources, and labor.

It seems like so many of us just want to go back to the way it was before. Not before, as in a few years ago, but a few generations ago. So far back that we ourselves didn’t even experience it. Fewer and fewer of my friends dream about becoming CEO of a tech company or running a hedge fund. More and more seem to be talking about moving to the country and homesteading. As children and young adults, our elders pushed us to engage in the rat race. Those who resisted were perceived as being afraid of competition. Perhaps some were, but an increasing number of young people are opting out not necessarily because they feel they can’t win, but because they have stopped believing that the game is even worth winning. We look at the “winners” of the modern world, and they don’t appear to be particularly virtuous or fulfilled. The people that Times Magazine tells us are the winners look more like the losers. The winners appear to be the ones who have opted out completely. Those who live simple, self-sufficient lives have hard days, I am sure, but they seem to have a purpose that cannot be replicated in a cubicle.

If I was the US treasury and knew that Bitcoin would inevitably replace my global fiat network, I would buy Bitcoin like a madman to hedge myself, but make sure to do it through the private sector to avoid a panic. This is what I believe MSTR is.

at some point, we need to have a serious conversation about whether Ted Kaczynski was right about tech.

An optimal AI setup should balance privacy and efficiency by routing all tasks through a local model first. That model would include a built-in understanding of its own limitations. When a request exceeds its capabilities, it could either delegate to a cloud model or, if privacy is a concern, ask for explicit permission before doing so. This kind of layered approach—rather than blindly sending everything to a single provider—is likely the future of AI usage.