Playtest Report — Round 2 — Corwin (MUD Veteran)

Resolution Audit — My Round 1 Top 5

#	Issue	Round 2 Response	My Verdict	Notes
1	Procedural Wilderness Persistence Are Not Specified — Where does cleared-dungeon state live? Shared world or per-player save?	New subsection §4 “World Persistence Model” commits: dungeon-cleared state is GLOBAL (stored in Universal Deck State’s `cartridge_history` bitfield). All operators see cleared dungeons. When two operators unlink, both see both operators’ clears (union merge).	RESOLVED	This is exactly what I asked for. The fiction is now committed: shared world, not per-player ghosts. Treasure stays cleared forever. That’s a MUD.
2	Hybrid Async/Sync Multiplayer Model Has Unspecified UX and Edge Cases — What happens when operators are in different zones during combat/puzzles?	New subsection §7 “Multiplayer Sync Protocol” specifies: solo combat can trigger in linked instances (screen shows “COMBAT: Solo mode”). Puzzles requiring both block solo (“LOCKED: Requires linked partner”). Cable disconnect has retry/solo-degrade/abandon options. Reputation changes during solo combat are LOCAL (per-operator).	RESOLVED	Solid. The spec now has teeth. Three sync contexts (exploration async, combat sync-optional, puzzles sync-required) + explicit error recovery (retry, degrade, abandon). This addresses my fear of mysterious failures.
3	CONS Is Overloaded and Needs Context-Clear Feedback — Three unrelated domains use CONS. Which context is active?	New subsection §8.6 “Context-Aware Action Hints” specifies on-screen hints change per domain: “[COMBINE: merge items]” in inventory, “[COMBINE: prepare offer to QUEST-GIVER]” in front of NPC, “[COMBINE: secure rope to hook]” in dungeon. UI disambiguates without renaming keys.	RESOLVED	Smart move. Keeps Lisp grammar unified (CONS is always construction) but makes the domain explicit on-screen. The spec shows three detailed examples. This will teach players the pattern quickly.
4	Knowledge System Is Invisible and Its Progression Is Unclear — How do you see what you’ve learned? How does it affect gameplay?	New subsection §6 “Knowledge and Learning Progression” specifies: Knowledge lives in `knowledge_index`. On acquisition, Cipher narrates: “KNOWLEDGE ACQUIRED: Cipher rotation pattern.” SYS menu includes “KNOWLEDGE” submenu (searchable list). Mechanic: Each learned fact grants +10% speed for related puzzles OR unlocks bypass.	RESOLVED	Excellent. Knowledge is no longer invisible. Cipher narration at acquisition, SYS menu to review, explicit mechanical effect (+10% speed or bypass). This is progression players will feel.
5	Inventory Combining Needs Uncombination and Puzzle Constraints — Can you uncombine items? What if you want rope AND hook separately?	New subsection §6 “Item Combination and Decombination” specifies: CONS creates new item from two originals. Original items are consumed (gone). Example: Rope + Hook → Grappling Hook (rope & hook disappear). To get rope back: find another rope OR CONS Grappling Hook + Scissors to reverse. Puzzle consequence: if you need rope AND hook separately, you need two copies.	RESOLVED	I asked for clarity on reversibility. Got it. Items are consumed on combination. Decombination is possible (costs Scissors). This creates real inventory-management puzzles. Elegant.

New Issues Introduced

None that are showstoppers. However:

The audio narration system is ambitious and costly. Section 9 now specifies a comprehensive audio-only play mode: every room narrated by Cipher voice (3 seconds each), NPC dialogue enumerated (~1 second per option), combat status updates every 3 seconds, etc. This is ~35–45 handcrafted room narrations + templates for procedural rooms. For ICE Breaker, this was a core feature. For Threshold, it’s positioned as essential for accessibility.

Assessment: Not a regression. It’s honest about the scope. But if this slips, the spec claims something the code won’t deliver. Keep this on the critical-path checklist for Round 3.

The onboarding sequence is scripted but not tested. Section 5 “Onboarding and Early Coop Hook” details a 30-minute arc: boot → link → tutorial NPC (KEEPER) → independent exploration → dual-lock puzzle at minute 15 (the “first coop moment”). This is clever design, but it assumes the KEEPER NPC is charismatic and the dual-lock puzzle feels organic, not forced.

Assessment: Not a regression. This needs playtesting feedback on emotional resonance. If the arc lands, it’s a huge win. If it feels scripted, it’s a huge miss. Worth the risk.

EXAMINE is a new verb, and key binding is deferred. Section 3 introduces EXAMINE as a dedicated verb for object inspection (Vex’s issue #2). Key binding is a placeholder: “candidates: F10 or shifted key. This costs a design decision (which key?).”

Assessment: Not a regression for gameplay, but it’s an implementation TODO. Needs to be resolved before code phase. The design is sound; the key assignment is bureaucratic.

What Got Better

The spec is now specific. Round 1 left gaps; Round 2 fills them. Every major system now has explicit rules: persistence model (global), sync protocol (three contexts + error recovery), knowledge feedback (narration + menu + mechanical effect), item combination (consume originals, reverse with Scissors). This is spec maturity.

On-screen verb hints are brilliant. Mae flagged Lisp jargon as a barrier. Round 2’s solution (on-screen “[ENTER: drill in]” instead of “[CAR: enter]”) teaches the paradigm while hiding the jargon. The keycaps still say CAR, but players never see it on-screen. This is the right balance between paradigm coherence and UX clarity.

The early coop hook (minute 15) is essential to multiplayer-first design. Round 1 promised multiplayer-first but had no coop moment until the 45-minute Obsidian Tower. Round 2’s dual-lock puzzle at minute 15 forces cooperation early. New operators feel essential by minute 20. This is the difference between “multiplayer is optional” and “multiplayer is the game.”

Multiplayer sync protocol is thorough. The hybrid async/sync model is complex, but Round 2’s spec details all three contexts (exploration async, combat sync-optional, puzzles sync-required) + cable-disconnect scenarios (retry, degrade to solo variant, abandon). This is playable, not theoretical.

Audio narration commitment is honest. Rather than hand-waving screen-off play, Round 2 spec details how every system (room descriptions, NPC dialogue, combat, inventory, map) has an audio equivalent. This is a big lift, but it’s transparent about what accessibility requires.

What’s Still Rough But I Can Live With

Procedural wilderness repetition after 10+ hours is still a risk. Round 2 flags it (section 12, “Procedural wilderness becomes tedious”) but doesn’t detail the solution. The spec says “vary biome structure” and “seed biomes with handcrafted encounter tables,” but doesn’t show examples. I trust the design team will handle this in code, but it’s not specified yet.

Sync failure recovery is clear, but latency tolerance is tight. The spec says latency < 200ms is “sync OK,” 200–500ms is “degraded,” > 500ms is “unstable, check cable.” For a TRRS serial link at 115200 baud over 1–3 meters, this is reasonable. But if a cartridge is at the edge of spec (long cable, noisy power supply), puzzles will fail mysteriously. This is infrastructure risk, not design risk. Acceptable.

Solo mode depth is still unproven. The spec says solo is “satisfying for 30–50 hours” (section 7, “Feel”) and represents “70% of multiplayer depth.” But we haven’t played it solo for 20 hours yet. Pressure-plate puzzles become “time-limit variants” (single-operator, 15-second clock). That could feel narrow. Playtest this aggressively.

The Cipher voice characterization is still minimal. Section 9 says Cipher always uses “880 Hz, professional tone.” But when does Cipher speak beyond narration? Does it comment on player choices? Does it have personality, or is it a purely functional narrator? Round 1 asked this. Round 2 improved the narration but didn’t deepen Cipher’s personality. That’s fine—it’s not a blocker—but it’s a missed opportunity for diegetic flavor.

Classic MUD Scorecard — Round 2

(Updating from Round 1.)

Room descriptions with personality: 5/5 → 5/5 (unchanged). Section 8 shows the same Market Square example. Cipher narration (Round 2) adds audio texture. Still the gold standard.

Exits that reward exploration: 4/5 → 5/5 (improved). Section 4’s explicit room hierarchy (Market Square → level 1 hubs → level 2 interiors → level 3 private rooms) makes the spatial structure learnable. CDR rules are now deterministic and intuitive. Exploring feels less random.

Inventory that matters: 4/5 → 5/5 (improved). Item combination now has irreversibility (CONS consumes originals) + reversibility (CONS Grappling Hook + Scissors to un-combine). This creates real inventory-management puzzles. Rope vs. Hook vs. Grappling Hook are strategic choices, not just convenience.

Combat with real decisions: 3/5 → 3/5 (unchanged). Still 8-second timer, action costs, accuracy model mentioned but not detailed. Round 2 didn’t deepen this. It’s solid but not a standout. I’d still like to see the full combat matrix (stat → accuracy, damage scaling, status effects). Playtesting will validate.

NPCs with character: 4/5 → 4/5 (unchanged). Same named NPCs (KEEPER, QUEST-GIVER, etc.). Relationship state persists. Dialogue trees are mentioned but not fully scripted. The spec is good but doesn’t showcase NPC depth. Round 2 didn’t change this, and I don’t fault them—it’s design-complete, just not showcase-polished.

Progression curve (first hour → hundredth hour): 4/5 → 5/5 (improved). Round 1 had implicit progression (reputation, knowledge). Round 2 makes it explicit: Cipher narrates knowledge gains, SYS menu shows your knowledge, each fact grants +10% speed or a bypass. By hour 100, players will know they’re powerful because the game tells them. This is the transparency I flagged in Round 1.

World persistence: 3/5 → 5/5 (improved). Round 1 was vague (is cleared-dungeon state shared?). Round 2 commits: cleared dungeons are GLOBAL, all operators see the same clears, the world is genuinely shared. This is no longer ambiguous. This is a MUD.

Player economy: 4/5 → 4/5 (unchanged). Credits, reputation, NPC trades, quest rewards. The economy self-corrects (inflation scaling). Round 2 didn’t deepen this, and I don’t expect it—it’s solid in Round 1. Still no player-to-player trading (operators can’t trade items with each other), but that’s probably fine for a two-operator game.

Fiction of “connecting to a server”: 5/5 → 5/5 (unchanged). “You are connecting to THRESHOLD, a persistent network.” This is still the soul of the design. Round 2 didn’t change it; doesn’t need to.

Mortality / stakes / consequence: 3/5 → 3/5 (unchanged). Death costs credits and reputation. Creatures remember being defeated. This is soft consequence, Aardwolf-style. Round 2 didn’t deepen consequence mechanics. It’s fine—not every MUD needs permadeath—but the design philosophy is “play without fear,” which may lower long-term tension. Playtesting will tell.

Final Call

SHIP IT WITH NOTES

Here’s why:

The core design is sound and the gaps are closed. My Round 1 Top 5 are all resolved or made clear. The spec is specific enough to implement. The audio narration system is ambitious but detailed. The early coop hook is clever. The multiplayer sync protocol is thorough.

But three things need explicit playtesting focus (not blockers, just high-risk):

Does the onboarding arc land emotionally? The dual-lock puzzle at minute 15 is supposed to feel like “we did something together that neither of us could do alone.” If it feels forced or clunky, the multiplayer-first promise falls apart. Run this with at least three new-operator pairs.
Does procedural wilderness stay fresh after 10 hours? The spec flags this as a risk. Run a solo campaign for 20+ hours in wilderness. Document when (or if) tedium sets in. If the 150th procedural room feels like a copy-paste, the world loses texture.
Does the knowledge system feel rewarding? Cipher narrates “KNOWLEDGE ACQUIRED” and shows a SYS menu entry. This is explicit, but is it satisfying? Some players need level-up fanfare. Others prefer quiet progression. Playtesting will show which this design is.
(Bonus) Verify sync protocol on actual hardware. The 200–500ms latency tolerance is reasonable for TRRS cable, but actual latency depends on cable length and quality. Stress-test with intentional cable degradation (loose connector, longer cable, noisy environment).

Resolution Summary

Corwin #1 (Procedural persistence): RESOLVED
Corwin #2 (Async/sync UX): RESOLVED
Corwin #3 (CONS overloading): RESOLVED
Corwin #4 (Knowledge invisibility): RESOLVED
Corwin #5 (Item combination): RESOLVED

4 of 5 resolved completely. 1 partial (Corwin #2 is thorough, but real-world hardware playtesting is needed to prove it works).

New issues: None that are design failures. Audio narration is ambitious; onboarding needs emotional validation; procedural wilderness tedium is flagged but not solved (by design—playtesting will guide the solution).

What Impressed Me This Round

The team listened. All five of my major issues got thoughtful responses. Some got simple fixes (knowledge menu). Some got detailed specs (sync protocol). Some got design changes (on-screen verb hints). Nobody punted on hard problems or added more complexity without addressing the original gaps.

The onboarding arc is novel. Most games tutorial-dump the controls. Threshold teaches verbs in context (KEEPER NPC as a diegetic tutor), then immediately uses them (independent exploration), then forces cooperation (dual-lock puzzle). By minute 20, a new operator has felt all three: “I learned the interface. I can navigate alone. I’m better with a partner.” That’s good game design.

The spec maturity improved dramatically. Round 1 had good ideas and gaps. Round 2 has good ideas and answers. This is ready to hand to a programmer.

Tone Shift from Round 1

I went in skeptical of Round 1. I left impressed. Round 2 is a genuine response to feedback—not defensive, not over-complicated, just “we heard you, here’s how we fixed it.” I’m less skeptical now. More cautious, maybe. The design works in theory. Playtesting will tell if it works in practice.

Would I still play this? Yes. Even more so now. The early coop hook sold me.

Corwin
MUD veteran (LPMud, DikuMUD, Achaea, Aardwolf)
Playtest date: April 21, 2026
Review session: 2-hour detailed spec audit + comparative analysis (Round 1 to Round 2)

Summary for the Team

Final Verdict: SHIP IT WITH NOTES. All major Round 1 issues resolved. The spec is now detailed and playable. Three high-risk items need playtesting validation (onboarding emotional resonance, procedural freshness after 10+ hours, knowledge system satisfaction). Audio narration is ambitious but detailed. No design failures, just implementation risk.

Resolution count: 4 of 5 completely resolved, 1 partial (sync protocol is thorough, but hardware validation needed).

New issues: None critical. Onboarding and procedural wilderness are flagged for playtesting, not design failure.

Report location: /sessions/wonderful-tender-faraday/mnt/kinoshita/docs/gameplay-specs/drafts/playtest-round2-corwin.md