Every methodology describes its limits from the inside. But when multiple methodologies are placed on a single map, a different layer becomes visible: a class of questions the field currently has no language to ask.
One of those blind spots concerns contradictions. Existing methodologies rarely distinguish between contradictions that should be resolved and those that should be maintained. The field often resolves tensions that may actually be structurally productive.
This observation led to StratoAtlas and CDSA — not as frameworks invented to fill a gap, but as the natural next question once the gap becomes visible.
The investigation runs through a trialogue: human researcher + two AI positions whose divergence produces a usable gap. That gap — not additive coverage, but the space between two irreconcilable contexts — is what makes the frame visible and opens access to the positional level.
MCP solved the infrastructure problem for dialogue: connecting agents to external systems safely. But trialogue requires structurally different infrastructure.
Dialogue accumulates context within a session. Trialogue requires state to persist across sessions and across agents. Without that persistence, the gap collapses back into dialogue.
This is not a feature request. It is an observation about where the trajectory leads — and about a structural gap in what is currently being built.
The case is documented. The observation is published:
stratoatlas.com/research/case-a-int-2026-001 →
stratoatlas.com/research/observation-003 →
If this distinction is relevant to the evolution of Claude's research workflows, I would be glad to continue the conversation.