09 · Experiments
Experiments & Prototypes
A working catalog of prototypes I use to test mental models before committing to full product designs.
By John Wright-Nyingifa · Product Designer building infrastructure for DeFi, DePIN, and autonomous agents.
Each experiment is designed to answer a specific question: Can users predict what the system will do? Can they understand delays and trust boundaries? Can they recover from failure without panic?
Active Experiments
Testing: how plans degrade under uncertainty, when agents should pause vs retry, how much autonomy feels comfortable. Prototypes: "step-through" simulator with phase states, policy sliders (risk, budget, frequency).
Testing: phase-based progress indicators, user comprehension of "included vs verified vs finalized." Prototypes: multi-hop timeline view, bridge + swap + deposit preview.
Testing: whether users can map delays to a layer (sequencer, DA, proving, settlement). Prototypes: modular "x-ray" status strip, dependency graph for a single action.
Testing: expressing constraints without jargon, showing tradeoffs (price vs speed vs trust). Prototypes: intent form with "guardrails," preview that includes worst-case bounds.
Testing: recovery paths that are explicit and safe. Prototypes: "Claim refund" flows, reroute suggestions with clear tradeoffs.
Testing: logs as the primary UX surface, storytelling that reduces support tickets. Prototypes: English-first log with expandable trace, counterfactual panel: "I chose X over Y because…"
Testing: whether users can trust a route they didn't choose. Prototypes: route cards with phases + assumptions, trust tier labels (Standard / Secure).
Testing: "what changed now" vs "what remains." Prototypes: before/after state diff block, checkpoint markers (DA published, proof verified).
Testing: making DA delays legible without teaching DA. Prototypes: availability confidence bar, "publishing backlog" indicator.
Testing: user perception of congestion and fairness. Prototypes: inclusion time distribution, priority lane vs standard lane comparison.
Testing: debugging cross-chain messages. Prototypes: message ID → lifecycle trace, source event → destination execution mapping.
Evaluation Framework
Four questions every prototype must answer:
Current state must be unambiguous.
Next phase must be predictable.
Recovery path must be visible.
Ranges and phases, not false precision.
Output Artifacts
High-fidelity mockups for each experiment.
Phase diagrams showing all possible states and transitions.
Status text and error messages for each failure mode.
Small clickable prototypes testing specific flows.