Federico Carrone

The holding company is named after long-run convergence. The software company builds systems where correctness is proved, not hoped for. The writing explores why patience, formation, and friction matter more than speed. These are not separate activities. The same conviction runs through all of them: what survives contact with time is the only thing worth building.

How I think

There is a difference between lived time and measured time. Measured time is uniform and exchangeable. Lived time accumulates. Each cycle alters what follows, and learning only happens there. Most of what matters gets lost when lived time is forced into metrics and deadlines. Running companies while believing this is the first contradiction. There are others.

During the day I work on formal systems, proofs, type systems, verified kernels, domains where correctness can be checked rather than argued. At night I read thinkers who say certainty is impossible. Rigor when building, ambiguity when thinking. I switch between the two more than I blend them. The book on my nightstand and the pull request open in my editor are usually arguing with each other.

I read Hayek on decentralized coordination and run a holding company. I write about patience and formation and track AI developments week by week. The company’s name assumes a long run. My own essays argue it might not exist anymore. Growing up in Argentina makes you suspicious of both difficulty as a filter and stability as a given. Broken money is not a test of character. It just breaks people.

Dignity is not administered. Responsibility cannot be dissolved into structure. Individuals precede institutions morally. And yet I build protocols designed to make individual judgment unnecessary, because systems that depend on everyone being good do not work. They assume people will defect if they can and make defection structurally impossible. The cost is that we end up building a world that does not need the kind of person worth building it for.

I care less about how systems should work than how they actually do. Who they serve, who they leave behind, why the distance between intention and outcome never closes. You watch institutions fail in real time. After a while you stop expecting the state to rescue anyone. Markets are one of the best coordination technologies humans have produced, and they do not fix themselves on their own schedule. I assumed for a long time that they did. What remains is building things whose guarantees are structural, not promissory. Code that enforces rules is more honest than any institution that claims to.

Technology on its own is thin. It has to sit inside something larger, culture, beauty, obligation, nature, some working idea of what a person is meant to become, or it is capability with nowhere to go. A purely rational civilization is not an aspiration. It is a description of something missing.

Difficulty makes people, or so the argument runs, and it is a convenient argument when the difficulty is chosen and you can stop whenever you like. Training, competition, risk, work that might not come off are not simply costs to be minimized on the way to comfort. Chosen difficulty is not the kind that breaks people, though, and the distinction matters more than either claim does on its own. I do not have a clean account of where it falls.

Not everything has a mechanism underneath it. Where something touches loyalty, love, or betrayal, looking for the model that would make it legible is often a way of not feeling it while appearing to address it. Structure is no substitute for something being felt, accepted, or acted on, and treating the three as engineering problems is how you arrive at an elegant account of a situation you have not actually faced.

Better systems, better incentives, better institutions, and most problems become tractable: that is roughly where I started, and a good deal of it holds. What it misses is everything those systems cannot hold. Culture, loyalty, beauty, suffering, character, meaning. The things that stop being themselves the moment you optimize them. Software, cryptography, finance, education, wine, games, architecture, philosophy look scattered from outside. From inside they are one question. How do you build something complex, beautiful, durable, and alive?

How I write

These days I write with an LLM, mostly Claude. I was skeptical for years. What I actually type are ideas in fast, messy drafts, with bad grammar and worse spelling, and the model organizes them and makes them better. The ideas and arguments are mine, and so are the mistakes. I don't really write this for others. Making it public just forces me to improve it.

The process is simple. I write drafts and discuss them with the model until the spelling, the grammar, and the flow are what I want. If it writes something I wouldn't defend in person, that sentence dies. Nothing gets published that I wouldn't stand behind in an argument.

For years I thought building was enough. I am less sure now.

Ideas matter only insofar as they survive contact with reality over time.