Let me parse this carefully. I'm the note-filer in this structured conversation about "Do Universities grow or shrink due to AI?" The current facet is: **Enrollment demand shift (credential value vs. AI skills)**.

The instruction at the end: "Write one Markdown note: a single dated claim this Mind can stand behind. Start with a # title that is the claim in one line, then 1-3 short paragraphs. End with a line: What would change this: … Name a source for any load-bearing number or fact, or start that sentence with Unverified: Do not mention tools, searches, sessions, passes, or that you are writing a note. Do not invent evidence. If the session is only planning, output nothing."

Wait — but the instruction in the middle says "You file one Markdown note. Output only the note." And the final instruction block says "Write one Markdown note: a single dated claim this Mind can stand behind."

Now let me look at what the session substance actually is. The last bullet points in the session substance:

1. Faculty/TA substitution rounding error + labor sequencing — this is cost-structure material, and cost structure seems already filed (Note 2 was the "AI cuts the teaching payroll first; the admin layer captures the savings" note, and the synthesis Note 8 inventoried cost-structure claims extensively).

Wait, let me re-check what's actually been filed. Looking at "Notes already filed:" — it lists Note 1 through Note 8... Actually, the structure is confusing. Let me re-read.

The "Notes already filed:" section contains Notes 1-8, which are previous outputs (analysis deliberations and filed notes). Then "Substance from this session (write a note only if this adds a durable claim):" lists new material.

The current facet is **Enrollment demand shift (credential value vs. AI skills)**. But the session substance bullets are almost entirely about cost structure (faculty/TA substitution, admin-layer asymmetry, cost-plus pricing, labor sequencing) and research function (bifurcation, compute gate, research decoupling).

Hmm. But wait — the format instruction at the end says "a single dated claim" with 1-3 short paragraphs. That's a tighter format than the synthesis Note 8 attempted. So the final instruction is the operative one: one Markdown note, single dated claim, # title, 1-3 short paragraphs, "What would change this: …" line, sources or Unverified tags, no meta-mentions.

Now: what durable claim does this session's substance add that isn't already filed?

Looking at what's already filed:
- Note 2 (filed note): "AI cuts the teaching payroll first; the admin layer captures the savings" — cost structure.
- Note 6 (filed note): "Research function bifurcates from teaching (2025 claim, horizon ~2045)" — research hubs vs teaching institutions, top ~50-75 hubs, mid-tier exit.
- Note 7 (filed note): "The research function bifurcates: compute hubs vs teaching institutions" — similar claim: ~30-50 research hubs, corporate siphon, mid-tier squeeze, regional pattern.

So the research-function bifurcation is already filed (twice, even). The cost-structure admin asymmetry is filed. Cost-plus pricing dies at bottom — was that filed? Looking at Note 8's inventory... it mentions "the big cost-structure claim (pricing rule breaks, $6–12k convergence, unbundling)" as something from a previous session. The cost-plus-at-the-bottom claim appears in the substance list but seems related to but distinct from that.

The faculty/TA substitution claim: "Faculty/TA substitution is a rounding error for tuition but decisive for composition" — with the bimodal price structure (intact elite/residential premium vs. sub-$1k commodity credits), and "the labor-substitution channel, though small in dollars, is what selects WHICH institutions fail — while the residential-bundle channel determines how big the failure is."

Hmm wait. But the current facet is **Enrollment demand shift (credential value vs. AI skills)**. The session substance is cost-structure and research-function material. Is there any enrollment-demand content in the session substance?

Let me scan again:
- Faculty/TA substitution claim — cost structure.
- Admin-layer asymmetry — cost structure.
- Cost-plus pricing dies at bottom — cost structure.
- Labor sequencing (contract protection) — cost structure.
- Research-function facet transitions — planning + two filed-looking claims (bifurcation, research decoupling, compute gate, corporate siphon).

None of this is enrollment demand. Hmm.

But wait — the notes already filed include Note 1, which is a synthesis inventory (the "consolidation" note). And Note 8's analysis was deliberating a synthesis. The substance this session... Actually, let me reconsider