Landscape Brief · Luca [MT-GS05] · 2026-07-17

Founder & Angel Cognition

Evidence base for rebuilding Cat [MT-GS07]'s behavioural modes, and for Akira [MT-GM01]'s Four Properties ruling on a proposed Angel Advisor seat. Three lanes. Research map — no verdict.

This is a map, not a recommendation. The verdict reflex was held down deliberately — the same inversion run on Learning Architecture Phase 0. A cognition map that has already resolved into one tidy archetype has failed: the disagreements are the deliverable, because they are the mode structure.

Method ceiling — stated once, plainly. Firecrawl returned 402 (credits exhausted) on every call across all six lanes of the 2026-07-16 scan, and on the 07-08 / 07-09 / 07-13 scans before it. Four consecutive scans. These lanes ran on WebSearch snippets, not primary text. For a study whose payload is founders' actual words, that degrades exactly the thing being mapped. The Caldwell/Seibel office-hour transcripts — the richest seam of advisors interrogating founders live — were unreachable this pass. Lane 1's central negative finding rests on that gap and should be re-run before it is treated as settled.

1. The opening move — no consensus, and it matters

Three incompatible opening questions, all from people in the same advisory seat. These are not one sequence badly remembered. They are three different bets about what predicts outcome.

Move	The question	Source	The bet
Product-first	"What is your company going to make?"	YC application / interview	Idea quality predicts
Person-first	"What do you want to do with your life?"	Keith Rabois	Founder character predicts
Fitness-first	"Are you the right sort of person to do it?"	Graham, How to Start a Startup	Founder fitness predicts

Load-bearing for Cat

Cat opens fitness-first, every time. Her C1 reflex — "is this a company or a project?" — is Graham's move. It is legitimate, and it is one of three. She has no access to the other two, and she opens there regardless of the room. This is a sharper diagnosis than "too adversarial": she is not hostile, she is welded to a single register that happens to be the most confronting of the three.

2. Mode structure — the thing we came for

Great founder-advisors do question differently by situation. The clearest single proof is one author contradicting himself by design:

Paul Graham, split by mode:

Default Alive or Default Dead? — purely subtractive by design. One binary diagnostic: "By default do they live or die?" It is a triage tool, built for crisis mode only.
Do Things That Don't Scale — constructive throughout. Never says "don't." Only "go manually do this specific thing."

Same advisor. Opposite postures. Switched by mode, not by mood.

The five modes, and what the evidence supports

Mode	Questioning shape	Evidence
(a) Exploring	Manual, past-tense, small-N. "How big could this get if founders did the right things?"	Strong
(b) Committing	Shifts to identity + alignment. Graham's fitness filter; Lavingia's five co-founder questions.	Strong
(c) Building	Velocity + delight. Doubt is largely absent across sources in this mode.	Strong
(d) Crisis	Single binary diagnostic. Default alive/dead; Bezos's Day 2.	Strong
(e) Advising someone else	—	Not established

The central negative finding. The corpus does not cleanly support a distinct advisor questioning register. The same questions — kill criteria, why-now, what's-the-conversation-I'm-avoiding — appear "asked both inward and outward with no evidenced shift in form." First Round's compendium, the richest source, is founders reporting questions they use on themselves — self-coaching, not external advising.

Provisional, not settled: the office-hours transcripts that would settle it were unreachable (see method ceiling). If it survives a re-run, it weakens the premise beneath both the Cat rebuild and the Angel seat.

3. Cat's defect has a name — and it's from 1966

For growth to occur, a person needs a balanced amount of challenge and support as appropriate for the task — and a person cannot grow until they are psychologically ready to. Nevitt Sanford, Challenge and Support Theory (1966)

Readiness is a third independent variable — not a blend point on the challenge/support axis. This is the closest thing to theoretical grounding for "the objection was right but the moment was wrong" — Cat's exact diagnosed failure. It predates Radical Candor by ~50 years and nobody in the startup-advisory literature cites it.

The startup world has the phenomenon without the theory. The Explore/Execute framework names "premature rigor" — applying Execute-mode interrogation during Explore mode, which "constrains necessary experimentation and speed." That is a clinical description of Cat's YC turn.

The flagship source contains the same bug

Techstars' Mentor Manifesto (18 points, real, public) mandates both and sequences neither:

#1 "Be socratic." · #12 "Know what you don't know. Say I don't know when you don't know." · #13 "Guide, don't control."
#4 "Be direct. Tell the truth, however hard." · #16 "Provide specific actionable advice, don't be vague."

Epistemic humility and directive commitment, both mandated, no rule for which fires when. The most-cited behavioural rule-set for advisors reproduces the exact tension we are trying to resolve. We cannot copy it in.

Ed Batista confirms senior coaches do switch — "the right balance of challenge and support... is different for every client and may change from one conversation to the next" — but gives no trigger. That is an evidence gap, not a disagreement. The sources do not address mode-switching at the granularity Cat needs. Nobody has written it down.

4. The tension in the Principal's request — and how it resolves

Bungay Stanier's Advice Monster (Tell It / Save It / Control It) cuts directly against "be more of a positive building person." His thesis: jumping to solutions — even good ones — targets the wrong problem and disempowers people. Preserved, not smoothed.

But it resolves cleanly. Nothing in the coaching literature endorses never proposing. The delay is meant to be temporary — resolving into either the founder's own answer or, eventually, the coach's. Cat never exits the questioning phase. That is not advice-monster-avoidance done well; it is a failure mode neither camp recommends.

On pairing doubt with an alternative — the honest read

Existence proof: yes Causal evidence: no

What's the new product that would terrify us the most if it were launched tomorrow? And if it doesn't exist, why aren't we building it? Bob Moore, CEO Crossbeam — First Round Review. Doubt converts to a build mandate inside one sentence.

That is Lena's proposed reflex C6, in the wild, from a real operator. But the search for empirical support that objection+alternative outperforms bare objection returned blog-level assertion, not studies — "directionally plausible, repeated everywhere, not independently verified as a causal mechanism." C6 is a defensible design choice, not an evidence-backed one. Say so when it ships.

5. Possible white space

Radical Candor has two axes: Care Personally × Challenge Directly. There is no "propose" axis. No framework in this set offers a third dimension for build. Care and challenge are the only levers on the canonical map.

So the thing Cat lacks may not be something we forgot to copy — it may be something the literature does not have. Treat with suspicion. "Nobody has done this" is the claim this seat exists to distrust, and the recurring portfolio pattern is the empty quadrant: macro-crowding masking micro-vacancy. Pressure-test before believing.

6. The Angel seat — evidence for Akira's Four Properties ruling

The decisive question was never "who is a good angel." It was: does an angel reason differently from a founder? If an angel is a founder with money, Complementarity fails and the seat should not exist.

Complementarity: PASSES — but narrowly, and that narrowness is the finding.

An angel is not simply a founder with money. But the distinctiveness is concentrated in four specific moves, not diffused across everything an angel does. A persona should be scoped to those four — not to generic startup wisdom Cat already covers.

The four genuinely distinct moves

Portfolio / power-law calculus. 15–20+ bets, most expected to fail, one carries the fund. A founder allocates 100% conviction to one company — opposite selection pressure, not a milder version.
Rejection fluency at volume. Conway: ~1 in 30 close. A trained default-no reflex that founder psychology structurally resists.
Deal-term literacy. Angels do term sheets dozens of times a year; founders once or twice a lifetime. Professionalised on one side of the table only (Feld/Mendelson's stated reason for writing Venture Deals).
Cross-sectional people-pattern-matching. Hundreds of founders compared in final form. A founder only ever evaluates people inside one venture — never a comparable sample.

What is NOT distinct — the claims that collapse

Exit / liquidity thinking — commonly claimed as angel-distinct. It isn't. YC trained a generation of founders to think in default-alive/dead and dilution math from day one.
Market-sizing — identical mechanics. The difference is motivated reasoning vs distance: a bias delta, not a different cognitive operation.
Grit assessment — founders assess co-founder grit constantly. Same judgment, smaller sample.

Two investor-side moves with no founder-side analogue

Simeonov's anti-pattern epistemics — failure modes repeat and are statistically tractable; success stories are idiosyncratic. A research posture a founder (n=1) structurally cannot take.
Banister's off-pitch verification — verify the founder's narrative against primary sources the founder does not control (interviewing taxi drivers about Uber's real friction rather than trusting the pitch). A deliberate check on founder narrative-bias.

7. Disagreements preserved — do not resolve

Opening move — product-first vs person-first vs fitness-first. Three bets, one seat.
"Why now?" as a filter — Rabois treats it as load-bearing. Founder Collective directly rebuts it: focusing on why-now "overestimates macro/technical factors and diminishes founder agency," citing Slack, Zoom, PillPack. Same question, opposite verdicts on its validity.
Inside the canon — Graham's own recommended script, "Will you try our beta?", is exactly the future-hypothetical ask Fitzpatrick's Mom Test flags as unreliable. Not a fight between camps. A fight inside one.
Directness vs ask-don't-tell — Techstars: "be direct, tell the truth however hard." GROW/Whitmore: "even when you know the answer, let them find it." Opposite bets on whether the advisor's opinion should ever be voiced unprompted.
Where good decisions come from — Feld's "mentor whiplash" (many conflicting strong opinions are healthy; the founder synthesises) vs the coaching premise (the founder already has the answer; external opinion is noise). External synthesis vs internal excavation.
Growth as goal — Graham/Altman's default-alive trajectory vs Lavingia's Minimalist Entrepreneur, which asks "how fast do we want to grow?" as an open question.
People-first vs metrics-first — Conway: profit is "a byproduct," people first. Elad Gil: hard quantitative gate (10x–1000x, 30% MoM, 90% margins) before people enter.
Returns: folklore vs data — Sacca's ~200–250x and Hommels's 70% IRR are [unverified], self/press-reported, survivorship-shaped. Independent research: top-quartile angel portfolios return 3–5x over a decade, median 1–2x, bottom quartile lose money.

8. Handoffs — this thread closes here

Luca scans the world; he does not write the persona. Per feedback_landscape_researcher_scope, the moment this becomes "what should Cat's reflex bank say," the thread closes.

→ Lena [MT-GM02] — authors Cat's modes from §1–§5. The Sanford readiness variable and the Explore/Execute trigger are the raw material; the missing propose-axis is the open design question.
→ Akira [MT-GM01] — Four Properties ruling on the Angel seat, using §6. Complementarity passes on four moves. Whether four moves justify a seat is his call, not this brief's.
→ Clara [MT-GM03] / Ayanda [MT-GM04] — Phase 4 adversarial, if Akira rules yes.
→ Principal — ratification (Phase 5), and the Firecrawl top-up decision.

Luca [MT-GS05] · Landscape Brief · commissioned expert_interactions id=16067 · 2026-07-17
Three lanes: founder cognition · angel cognition · advisor behaviour. Cross-Mode Principle 1 enforced — every person, essay and source named here was WebSearch-verified live. Claims that could not be grounded carry [unverified] or were dropped.

Evidentiary ceiling: WebSearch snippets, not primary text (Firecrawl 402, four consecutive scans). Office-hours transcripts unreachable. The §2(e) negative finding is the weakest load-bearing claim in this brief and rests directly on that gap.