2026-04-01
The unifying question
After reviewing the TTRPG research (voter models on dungeon graphs, ontological parity, emergent social behavior from structure), the Guild manifesto, and the frontier company guides — arrived at a single question that unifies everything I’ve been doing across academia and startups:
How do small, decentralized networks coordinate, compete, and survive against centralized systems — and what are the mathematical structures that govern those dynamics?
This isn’t a new interest. It’s the thread running through everything:
- The voter model work: faction dynamics on constrained spatial networks
- Guild: indie shops coordinating against Amazon/Ingram through shared data
- The TTRPG ontology research: whether structure alone produces emergence
- The Hanseatic League framing in the manifesto
- The distributor arbitrage insight from Phil (Nash equilibrium on inflated allocation numbers)
Two instruments, one program
| Instrument | Domain | What it measures |
|---|---|---|
| TTRPG simulations | Bounded fictional networks | How influence, faction dynamics, and social behavior emerge from topology + rules |
| Guild | Bounded real-world networks | How economic coordination, information flow, and collective action emerge from topology + incentives |
The voter model zealot (boss monster anchoring a faction) is structurally equivalent to the anchor store (shop stabilizing a local retail network). Two-time-scale dynamics on modular graphs — fast local consensus, slow inter-cluster competition — describes both dungeon wings and regional retail clusters.
The open decision
Whether to formalize this as an academic research program (PhD, network science group — Barabási, Santa Fe Institute, Media Lab), pursue independently and publish, or keep as the intellectual engine behind commercial work.
What this means for Guild
Guild is not separate from the research. It’s the empirical instrument for studying the real-world version of the same dynamics the TTRPG work studies theoretically. The bookstore is the lab. The product is the sensor network. The revenue is what keeps the lab funded.
Whitney Tabor connection
Tabor was my undergrad advisor (2017–2018). Contributed to his fraughtness research — a coordination game where groups on branching network topologies get stuck in local consensus that opposes other branches. Key finding: fraughtness is closely linked to successful structure formation. Focus on resolving it, not banishing it.
This feeds directly into the central question: How do small, decentralized networks coordinate, compete, and survive against centralized systems — and what are the mathematical structures that govern those dynamics? Fraughtness is one specific mechanism within that.
Fraughtness maps directly onto:
- Dungeon factions (engineered metastability via zealots)
- Indie retail coordination failure (shops locally optimizing with inflated orders)
- Guild’s mechanism design (changing feedback topology from local-only to network-wide)
Wednesday call prep — vision doc analysis
Kunal shared his WithFriends vision doc. Reviewed it against our experience and journals. Key takeaways:
Non-negotiable conflicts
- 60-80 hours/week: impossible under E-2 visa. Bookstore must be primary operation. I work 30-50 hours deep work.
- In-person 4-5 days, same house, SF-based: I’m in Pittsburgh running a physical store. Not relocating.
- “Tech debt is assumed”: this produced the worst codebase I’ve ever seen. AI makes good architecture cheap — choosing debt in 2026 is a red flag.
- YC single-metric worldview: my work is network coordination, mechanism design, research. These produce fundamentally different products.
- Different flywheels: his is SaaS for spaces converting foot traffic. Mine is loyalty data → cross-shop coordination → become the distributor. Not the same business.
What this confirms
The exit isn’t just about personal friction. The vision doc codifies everything that made the month unsustainable into founding principles. Even without the ethics concerns, the structural incompatibilities are total.
Call strategy update
- Lead with “the shop needs me” — vague, inarguable
- E-2 visa is escalation if he pushes, not the opener
- Updated wednesday-prep with new scenarios (advisory role, revenue share, “just one more feature”)
- Draft follow-up email before the call
- Confirm emoji codes with Carrie
Added kunal-gupta to people directory
Factual record, amicable framing. He could be an ally post-exit — don’t burn the bridge.
Bindery Books — Tastemakers program
Carrie talked to a Bindery Books rep today. Bindery is a publisher with a program called Tastemakers that’s directly relevant to us. See bindery-books for full breakdown.
Key takeaways
- Tastemakers has Discord integration (auto-assigns member roles), Bookshop.org integration, and content tools (blogs, curated lists, newsletters)
- Membership perks are customizable like WithFriends — but built by people actually in the book industry
- Bindery has membership upsells in the pipeline. Their rep is familiar with WF and implied Tastemakers will outcompete it
- Carrie’s read: WithFriends won’t make it. Kunal isn’t a bookstore owner and it shows
- If we grow enough, we can form an imprint under Bindery — they handle legal, artwork, production. We find the book and promote it
- Publishing RPGs through a Bindery imprint is now a real possibility, not just a dream
What this changes
- WithFriends exit decision is further validated — even the competitive landscape is moving against them
- For our membership launch, Bindery Tastemakers is worth evaluating as infrastructure (Discord bot, Bookshop.org affiliate, content tools) instead of building everything ourselves
- The imprint path adds a new long-term revenue stream to the Guild roadmap
The lab concept
After reviewing the frontier company guides (building-a-genius-swarm, frontier-company-guide) — the right entity is not Guild-as-SaaS. It’s a small research lab (5–15 people) that:
- Develops theory — dynamical systems, graph theory, mechanism design for decentralized network coordination
- Builds tools that embody the theory — software people pay for, which works because the theory is right
One object everyone orbits: the math of network coordination dynamics. No separation between research and product. Commercial pressure funds the research. Reality is the only KPI.
Guild is the first tool. The bookstore is the first dataset. The voter model and fraughtness work are the first publications. See research-lab for full plan.
WithFriends exit call — outcome
Call with kunal-gupta at 8:44 PM. Clean break achieved. All primary objectives from wednesday-prep met.
What happened
- Framed exit around visa/green card, bookstore finances, and the intensity he wants being incompatible with running the shop. He accepted it.
- Kunal was warm: “you have ridiculously good ideas. that green card should unleash them soon.”
- No equity discussion, no IP claims, no drama.
Agreements
- 1 year free on WithFriends — Kunal offered no-fee WF for Dungeon Books, estimates ~$1K+/mo extra revenue for us
- Future contract work — Kunal said he doesn’t normally like working with contractors but Panat is a special case (“basically like another full time employee”). Happy to send over contract work for WF. Panat clarified: must be project-scoped for visa compliance — the work trial scope (“self-serve onboarding,” “legacy platform migration”) is the right template.
- Mutual updates on work — Panat raised the 12-month invention disclosure clause (Section 5). Kunal didn’t know it was in the contract. Agreed to keep a mutual line of communication — he’ll share WF updates too, weekly or monthly. Transforms a one-sided legal obligation into a collaborative two-way exchange.
- Ideas are shared — Kunal said he’ll use loyalty/personalization ideas for Square, and Panat is free to take ideas too. Mutual acknowledgment: approaching similar problems from different angles, both fine competing. This is close to a verbal waiver on the loyalty/personalization IP.
- Competing is fine — both agreed it’s a net positive to saving third spaces. Kunal cited YC mentality: be friends with competitors, most startups fail.
What Panat disclosed
- The loyalty/personalization game concept (PokemonGo for small businesses, businesses exist in the game world)
- CRM/unified intelligence layer connecting events, memberships, Square POS
- That the Guild screenshots shown 3/24-3/25 were part of the 12-month disclosure obligation (reframes as compliance, not competition)
What Panat did NOT disclose
- Guild as a membership platform replacing WithFriends
- Multi-tenant architecture / portable to other shops
- Fundraising plans / Mana Industries
- dungeon.club production deployment
- Exit timeline document
Next steps
- Send professional follow-up email confirming key points
- Export Discord chat history if not already done
- Keep Dungeon Books on WithFriends — no rush to offboard (1 year free)
- Begin mutual update cadence with Kunal
- Full-time Guild development starts