[EPIC - First Experience] Join game recommendation in create-game

[JorenC]

PRD: Smart Game Lobby Recommendation (Existing Game Suggestion)

Summary

We see a lot of similar lobbies, likely created by new players following the 'create game' path, then abandoning.

When a player is creating a new game (during variant + time selection), the system should detect if a similar public game already exists and recommend joining it instead. This reduces lobby fragmentation, improves game start rates, and enhances the onboarding experience for new players.

The recommendation is shown inline in the game creation flow, before final confirmation (step 3), and highlights the closest-to-filling existing game.

---

Problem Statement

Currently, multiple near-identical lobbies are created:

• Fragmenting player base
• Slowing game start times
• Creating a poor first-time user experience
• Increasing abandonment of near-empty games

New players often do not realize an existing suitable game is already available.

---

Goals

• Increase game start rate speed
• Reduce duplicate / near-identical lobbies
• Improve beginner onboarding experience
• Encourage healthier lobby population distribution

---

Non-Goals

• Forcing users into existing games
• Removing ability to create custom games
• Changing matchmaking system fundamentally

---

User Story

As a new or casual player, I want to quickly join a game that will start soon, so I don’t have to wait in an empty or slow-filling lobby.

---

Feature Description

Trigger Point

When a user:

1. Selects:
   • Variant (e.g., Classic, variants)
   • Time format (e.g., 24h / Classical / fixed deadline)
2. AND the game is public (not private)

→ System checks for existing matching lobbies.

---

Matching Logic

A “similar game” is defined as:

Required Match Criteria

• Same variant
• Same game mode (e.g., 24h / classical / fixed deadline type)
• Public game only
• Game not started yet

Scoring / Ranking

Each candidate game gets a “fill score”:

• % of players filled (higher is better)
• Age of lobby (older but still active preferred less than actively filling ones)

Primary selection:

• Game closest to being full (highest player count relative to capacity)

---

Special Rules

1. Duration-Fixed Games (Always Recommend)

• If a duration-fixed game exists with matching settings → always show recommendation block
• These are considered highest priority due to unpredictable start behavior and no time zone restrictions.

2. Fixed-Time Deadline Games

Recommend if:

• Deadline difference is within ±3 hours (user-local time adjusted)
• Deadline converted to user’s timezone before comparison

---

UI / UX

Placement

• Shown inline on game creation page (no dialog or interruption)
• After variant + time selection
• Before final “Create Game” step (Step 3)

Component: Recommendation Block

Design:
A highlighted, non-intrusive banner/card.

Example copy:

⚡ A similar game is already filling

A public Classic 24h game is nearly full and will likely start sooner.
Joining it may get you into a game faster.

CTA Buttons:

• Primary: Join existing game
• Secondary: Continue creating new game

---

Behavior

Clicking Recommendation

• Redirects to existing game’s Game Info page

Dismissing Recommendation

• User can continue to advanced settings or proceed with creation

Advanced Tab Exception

Users who explicitly enter “Advanced Setup” bypass recommendation after optional initial display.

---

Edge Cases

• No matching game → no block shown
• Multiple matches → only top-ranked (closest-to-full) shown
• Private games → ignored entirely
• Dead/inactive lobbies → excluded (no activity threshold met)

[johnpooch]

One thing worth flagging up front: we've actually already built most of this. The create-game flow already detects similar pending games and offers to let you join one instead:

• When you create a public, duration-based game, we look for an existing pending game with the same variant and movement-phase duration. If we find one, we show a dialog — "A similar game is already forming — join it instead to start playing sooner" — with Continue (create yours anyway) vs Join Them? options.
• There's also a dedicated Find Games screen with variant + duration filters and a "Fastest Start" recommendation that surfaces the lobby closest to filling.

So the core idea is already live. The main differences from the PRD as written:

1. Timing — today the prompt fires when you hit submit, whereas the PRD suggests showing it earlier (right after variant + time, before the final step). That's a reasonable UX refinement, but it's a tweak to existing behaviour rather than a new capability.
2. Coverage — we currently only run the check for duration-based games and skip fixed-time games, so the ±3hr-window idea for fixed deadlines isn't implemented yet.

A couple of broader thoughts:

On priority. The harm from duplicate lobbies is real but fairly soft — a stale pending game just sits there; it doesn't break anything or block other players. Given the nudge already ships, I'd put further polish here below the things that affect live games.

On the simplest effective fix. The "recommend a similar game" approach is a nudge — dismissable by design (the PRD rightly lists "forcing users into existing games" as a non-goal). That reduces duplicates probabilistically, but it does little about the case that actually triggered this: a single user spinning up many lobbies in a row. A hard cap on the number of open/pending games one user can create targets that directly and is simpler to build — one backend validation and an error message, no new UI or matching heuristics to tune. The two aren't mutually exclusive, but if we only do one, the cap hits the root case with less surface area.

On duplicate names (David's point) — I don't think this one's worth chasing. Games are keyed by ID, not name, so duplicate names are harmless; enforcing uniqueness would add a validation/failure path for no real benefit.

Net: given the recommendation flow already exists, I'm inclined not to invest much more here right now. If we do act, I'd lean toward including fixed-time games and the per-user pending-game cap as the cheaper, more targeted lever than reworking the recommendation UI.

[JorenC]

Well - I didn't know we had this already. We can re-open it later.
I think this UX is better, but still. Let it goooo