Many years have passed since I read Snow Crash, a science fiction novel by Neal Stephenson. At the time, while I struggled with floppy disks, the dystopian world it portrayed felt comfortably distant.

The book, which famously coined the term Metaverse, described a universe in which language was executable, identity was fragmented across digital spaces, and people immersed themselves in virtual worlds through AI-driven systems. I remember being immersed in the imagery and possibilities it evoked—but not concerned. Whatever future the book proposed did not feel imminent.

I was wrong.

What stands out in retrospect is not that Snow Crash discussed the evolution of specific technologies, but that it anticipated the failure modes of technology sprawl: systems that scale faster than understanding, abstractions that detach from accountability, and technical power that outpaces the social structures meant to contain it.

Many of those technological and social dynamics are no longer speculative.

Today, we deploy models that interpret language, infer intent, and influence decisions in real systems—often without a clear, shared understanding of what they do, how they fail, or how to explain their behavior when it matters most. We deploy powerful systems almost immediately, frequently before we have developed the understanding or controls required to manage their impact.

This site exists to think through that gap.

I work primarily in security and applied AI, in environments where systems are already complex, incentives are misaligned, and “just experiment” is rarely an acceptable posture. In those settings, the interesting questions are no longer whether we can build something, but whether we understand its boundaries well enough to trust it—or constrain it—once it is deployed.

What Snow Crash captured, intentionally or not, is a recurring pattern: when systems become powerful enough, they stop being purely technical. They become linguistic, political, and behavioral. At that point, failures are rarely caused by a single bug or model choice. They emerge from how components interact, how humans interpret outputs, and how institutions absorb—or ignore—risk.

That is where most of my attention sits.