Iconundrum
A daily emoji rebus with fast, fair competition.
- Status
- Live
- Year
- 2026
- Role
- Founder — product, design, engineering
- Link
- Visit
- Stack
- ReactCloudflareTypescriptLLM
Overview
Iconundrum is a daily rebus game built around one small, repeatable interaction: decode an emoji clue, submit the answer, and compare the result. The product is intentionally lightweight so each puzzle feels quick to start, satisfying to solve, and easy to return to the next day.
I built the player experience in React and deployed it on Cloudflare Pages, alongside an admin portal that lets non-technical stakeholders create, preview, schedule, and manage the puzzle calendar.
- Daily
- Shared puzzle
- Timed
- Ranking design
- 2 days
- Content lookahead
- AI
- Fallback generation
Designing the daily loop
A daily game needs enough friction to make solving meaningful, but not enough to turn a small ritual into a task. Everyone receives the same puzzle, creating a natural point of comparison, while the clue, answer input, feedback, and result stay within one compact mobile-friendly flow.
The content workflow is equally important. Stakeholders can manage future puzzles without an engineering deployment, keeping editorial ownership close to the people creating the game.
A fair time-based leaderboard
A countdown from midnight would rank availability rather than solving ability, and a browser-only timer would be easy to manipulate. The proposed ranked attempt therefore starts with an explicit action that records a server timestamp before revealing the clue. The correct answer records the finish time.
Incorrect answers add a fixed time penalty, so rapid brute-force guessing is not the fastest strategy. Each account receives one ranked attempt per day, with ties resolved by fewer incorrect answers and then earlier completion. Replays remain available but do not alter the leaderboard.
Protecting the publishing cadence
The system checks the scheduled content queue with a rolling two-day lookahead. If there is no approved puzzle in that window, an AI fallback generates a candidate rebus and answer using the same content model as an editorial puzzle.
Stakeholder-authored puzzles remain the primary creative voice. AI acts as continuity infrastructure: it provides time for review or replacement and prevents the daily experience from reaching an empty state when the planned queue runs short.
Product tradeoffs
Server-authoritative timing, one ranked run, and answer penalties address the clearest leaderboard exploits without over-engineering a casual game. The two-day fallback makes a similar compromise: it creates a useful intervention window without filling the calendar far in advance with lower-confidence generated content.