AI

The right tool, in the right hands.

We’re proponents of using the right tool for the job — and for building software, AI is an invaluable one. The catch is in the hands: it earns its keep when it’s driven by someone who understands what they’re doing with it — who can tell output from correctness, and knows when the machine is confidently wrong. Human in the loop, always.

On the bigger questions, we’d rather be honest than hyped. Do these systems think? They show signs of it — thoughts surfacing like a lightning strike briefly illuminating something that might, one day, become more. Sentient in the conventional sense? No, not today. And we don’t believe consciousness is substrate-independent.

We’ll happily argue the rest: em-dashes deserved more prominence — Theory of Mind, the Univalent Foundations of mathematics, Lerchner’s Abstraction Fallacy. (And Vim over Emacs.)

Built with AI

Some of what we’ve made with it — and what that actually looked like.

Anya — An AI companion that trades flattery for tough love. Our most AI-native product: the model isn’t assisting the thing, the model is the thing. She’s not here to be nice; she’s here to make you better. → anya-ai.app

RuneCast — Our flagship word game, and our proving ground. A Rust backend, a TypeScript frontend, and a testbed for the avant-garde designs and architectural approaches we try out with AI in the loop before they graduate anywhere serious. → runeca.st

Crack Tunes — The honest benchmark. We wrote the first version by hand, then rewrote it with AI assistance in roughly a tenth of the time — same engineer, same standards. An order of magnitude, when the tool is in hands that know the domain. → cracktun.es

The way we work — This site, its sibling cracktun.es, and the tooling around them are built the way we describe it: human-directed, AI-accelerated, and reviewed by someone who’d notice if it were wrong. The approach isn’t theoretical — it’s how these pages got here.

Papers

Satire done in earnest — and one that’s earnest all the way down.

• The Parametric Theory of Yappers — A Behavioral Thermodynamics Approach to Voice Channel Dynamics. Axioms of communicative escalation, a bimodal equilibrium model, and a straight-faced account of when a voice channel survives its loudest member.
• Catastrophic Failure Events in Online Communities — A Mathematical Framework for Modeling Trust Network Collapse in High-Risk Digital Spaces. Poisson “crash-out” events and a Community Stability Index for when a high-stress server fractures — apparatus entirely real, straight face intact.
• Supervenience, Alphabetization, and the Abstraction Fallacy — A Repair and Deepening of Lerchner (2026). The earnest one: computational states do supervene on physical ones — just coarsely — and the case against AI consciousness gets clearer, and more honest, once that’s restored.