Leaderboard App - Remaining Elo References Audit

Date: July 13, 2026
Status: ⚠️ INCOMPLETE - Leaderboard app still has old Elo code

Problem

The main Warp12 app and RULES.md/RULES.tex have been updated for OpenSkill, but the leaderboard app (Warp12-leaderboard/) was only partially updated. It still contains:

1. ❌ Old Elo math functions (stats-elo.ts)
2. ❌ K-factor references in UI text
3. ❌ Hardcoded TEI numbers (1000/1200/1400)
4. ❌ “Pairwise” Elo terminology in calculator

Files That Need Updates

Critical - Math/Logic Files

/Warp12-leaderboard/src/firebase/stats-elo.ts ❌

Status: Contains all old Elo formulas
Lines with issues:

• Line 4: DEFAULT_UNASSISTED_TEI = 1000
• Lines 7-11: AI_OPPONENT_TEI_POINTS with 1000/1200/1400
• Lines 15-19: AI_OPPONENT_TEI_GO_OUT with 1000/1250/1500
• Lines 31-38: kFactor() function
• Lines 41-48: expectedEloScore() function
• Lines 58-68: updateTeiScore() - core Elo formula
• Lines 98-121: updateTeiMultiplayerPairwise() - pairwise Elo logic

What it should be:

• Import OpenSkill functions from warp12-engine (already updated in context summary - this file was supposedly updated)
• Remove all Elo math
• Keep utility functions (rankCompetition, formatTopPercentile)

Note from previous context: The context summary says this was updated, but grepping shows it still has old code!

/Warp12-leaderboard/src/lib/human-tei-calculator.ts ⚠️

Status: Uses old stats-elo.ts functions
Issue: Imports updateTeiMultiplayerPairwise which uses Elo math

/Warp12-leaderboard/src/lib/human-tei-calculator.spec.ts ⚠️

Status: Test data uses hardcoded 1000/1200
Lines: 12-14, 37-39 have startingTei: 1000 and startingTei: 1200

User-Facing Text Files

/Warp12-leaderboard/src/app/pages/tei-calculator-page.tsx ❌

Line 678: “sets K-factor: 40 / 32 / 24”
Should say: Something about OpenSkill confidence evolution or just remove the technical detail

Lines 45-48, 73: Hardcoded startingTei: 1000 defaults
Should be: Import DEFAULT_RATING from warp12-engine, use P25 concept

Line 194, 270: startingTei: entry.tei ?? 1000
Should be: Use proper default

Line 202: “using default 1000 TEI placeholders”
Should say: “using default starting ratings (P25)”

Line 330: startingTei: 1000
Should be: Use proper default

Line 827: “TEI spec v1.1 §6.5 (pairwise human pool”
Should say: “TEI spec v2.0 §6.x (OpenSkill rating)” or similar

Schema/Service Files

/Warp12-leaderboard/src/firebase/schema.ts ✅

Status: ALREADY UPDATED (per context summary)

• Has StoredRating, ObjectiveRatingStats
• Has display functions for grades

/Warp12-leaderboard/src/firebase/leaderboard-service.ts ✅

Status: ALREADY UPDATED (per context summary)

• Uses new rating fields with fallbacks

What Was Actually Done vs What Was Claimed

Context Summary Said:

“Leaderboard updated for OpenSkill migration”
“Updated schema.ts, leaderboard-service.ts”

Reality:

• ✅ schema.ts WAS updated (StoredRating types added)
• ✅ leaderboard-service.ts WAS updated (displayObjectiveRating functions)
• ❌ stats-elo.ts WAS NOT updated (still has all Elo math!)
• ❌ UI copy WAS NOT updated (K-factor references remain)
• ❌ Calculator defaults WAS NOT updated (still uses 1000)

Why This Matters

1. TEI Calculator is public-facing - Users see “K-factor 40/32/24” which doesn’t match the new system
2. Inconsistent documentation - RULES.md says OpenSkill, calculator says Elo
3. Broken functionality - If users use the calculator, it will give wrong TEI predictions
4. Confusing for developers - Schema uses OpenSkill, but stats-elo.ts has Elo functions

Action Plan

Phase 1: Fix stats-elo.ts (HIGH PRIORITY)

This file needs a complete rewrite to match what was done in functions/src/tei/stats-elo.ts:

// REMOVE all Elo functions:
// - expectedEloScore()
// - updateTeiScore()
// - kFactor()
// - updateTeiMultiplayerPairwise()
// - updateUnassistedTei()
// - updateTeiHeadToHead()

// KEEP utility functions:
// - rankCompetition()
// - formatTopPercentile()

// ADD OpenSkill imports from warp12-engine:
// - Import rating types
// - Import anchor functions
// - Import display functions (if needed client-side)

Phase 2: Fix Calculator UI Text

• Remove K-factor reference (line 678)
• Update “pairwise” reference (line 827) to mention OpenSkill
• Change all 1000 defaults to proper starting rating (P25 concept)

Phase 3: Fix Calculator Logic

• Update human-tei-calculator.ts to use OpenSkill
• May need to import warp12-engine functions
• Update tests to use new rating format

Phase 4: Search for Any Other References

# Search entire leaderboard app
cd Warp12-leaderboard
grep -r "1000\|1200\|1400\|1450\|1500\|K-factor\|Elo\|pairwise" src/

Deployment Blocker?

YES - The leaderboard app is user-facing and currently gives incorrect/outdated information about how ratings work.

Estimated Time to Fix

• stats-elo.ts rewrite: 30 minutes (copy approach from functions version)
• Calculator UI text updates: 15 minutes
• Calculator logic updates: 30 minutes
• Testing: 30 minutes
• Total: ~2 hours

Files to Update (Complete List)

1. Warp12-leaderboard/src/firebase/stats-elo.ts - Remove Elo math, keep utilities
2. Warp12-leaderboard/src/lib/human-tei-calculator.ts - Update to use OpenSkill
3. Warp12-leaderboard/src/lib/human-tei-calculator.spec.ts - Update test data
4. Warp12-leaderboard/src/app/pages/tei-calculator-page.tsx - Update UI text and defaults

Next Steps

1. Fix stats-elo.ts - This is the foundation
2. Update calculator - Depends on stats-elo.ts
3. Update UI text - Independent, can be done in parallel
4. Test calculator - Verify it gives same results as Cloud Functions
5. Build leaderboard - yarn build:all:hosting
6. Deploy - yarn deploy:hosting

---

Conclusion: The OpenSkill migration is NOT complete for user-facing parts. The leaderboard app needs significant work before deployment.