Principles & Operating Discipline
The shift from "fun side project" to "sustainable business" is a vulnerable moment — community attention, personal pride, and reputation all raise the stakes. These principles are the guardrails we hold ourselves to so ambition stays grounded and the project stays honest.
Operating principles
- Deterministic truth above all. The target product's entire value is exactness. Correctness, reproducibility and transparent traces beat speed and breadth.
- Scope discipline. Be the best at one thing — isolated, stateless EVM simulation and cryptographic primitives. Say no to archive-node territory,
solc, ERC-app-layer logic, and consensus-layer mechanics. (See boundaries.) - Frictionless for outsiders, rewarding for insiders. Never put the community token in the critical path of a paying agent; let it be a discount and a perk, not a gate.
- Cypherpunk character. Permissionless access, open standards, privacy-respecting, code-first — carried forward from the side-project era into the business.
- Ship the ugly PoC first. Prove the thesis with a working end-to-end round-trip before perfecting the pipeline.
Founder traps we watch for
The vision was pressure-tested against a set of recurring psychological/strategic traps. Keeping them visible is part of the discipline:
| Trap | The risk |
|---|---|
| Over- / under-promising | Hype beyond checked fundamentals, or playing it so safe the idea doesn't matter. |
| Undervaluing assets | "AI can do everything" fatalism — forgetting the stack + experience aren't replaceable overnight. |
| AI sycophancy | Asking suggestive questions and getting confident confirmation instead of neutral analysis. |
| Forgetting economics / tech | Treating monetization or feasibility as an afterthought. |
| The community trap | Letting token-community tempo dictate the engineering roadmap. |
| Audience identity crisis | Building in the uncanny valley between human-visual and machine-headless — solved by two legs, one engine. |
| The infinite-AI-leverage illusion | Assuming easy creation means easy maintenance; over-committing as a solo builder. |
| Solution looking for a problem | Building the API first and hunting for users later. |
| Agentic-UX blindspot | Designing for humans when the consumer is an LLM that will silently misuse a bad schema. |
| Automated-debt avalanche | Trusting the AI-managed fork pipeline's happy path until a silent logic error surfaces. |
| Perfect-protocol procrastination | Over-engineering before a single agent has queried it in the wild. |
These also make good public-thread material — the project documents its own reasoning in the open.
This page is a living checklist; refine as the project teaches us new lessons.